OpenAI Launches ChatGPT Agent: A New Era of AI-Powered Task Automation

ChatGPT's new AI agent can browse the web and create PowerPoint slideshows

On Thursday, OpenAI launched ChatGPT Agent, a new feature that lets the company's AI assistant complete multi-step tasks by controlling its own web browser. The update merges capabilities from OpenAI's earlier Operator tool and the Deep Research feature, allowing ChatGPT to navigate websites, run code, and create documents while users maintain control over the process. The feature marks OpenAI's latest entry into what the tech industry calls "agentic AI" -- systems that can take autonomous multi-step actions on behalf of the user. OpenAI says users can ask Agent to handle requests like assembling and purchasing a clothing outfit for a particular occasion, creating PowerPoint slide decks, planning meals, or updating financial spreadsheets with new data. The system uses a combination of web browsers, terminal access, and API connections to complete these tasks, including "ChatGPT Connectors" that integrate with apps like Gmail and GitHub. While using Agent, users watch a window inside the ChatGPT interface that shows all of the AI's actions taking place inside its own private sandbox. This sandbox features its own virtual operating system and web browser with access to the real Internet; it does not control your personal device. "ChatGPT carries out these tasks using its own virtual computer," OpenAI writes, "fluidly shifting between reasoning and action to handle complex workflows from start to finish, all based on your instructions." Like Operator before it, the agent feature requires user permission before taking certain actions with real-world consequences, such as making purchases. Users can interrupt tasks at any point, take control of the browser, or stop operations entirely. The system also includes a "Watch Mode" for tasks like sending emails that require active user oversight. Since Agent surpasses Operator in capability, OpenAI says the company's earlier Operator preview site will remain functional for a few more weeks before being shut down. Performance claims OpenAI's claims are one thing, but how well the company's new AI agent will actually complete multi-step tasks will vary wildly depending on the situation. That's because the AI model isn't a complete form of problem-solving intelligence, but rather a complex master imitator. It has some flexibility in piecing a scenario together but also many blind spots. OpenAI trained the agent (and its constituent components) using examples of computer usage and tool usage; whatever falls outside of the examples absorbed from training data will likely still prove difficult to accomplish. For example, the ChatGPT Agent System Card shows that the agent can fail at complex tasks that require chaining together many steps in a novel way. In a "Cyber Range" evaluation, the agent was tasked with conducting a full-scale operation in a simulated network designed to mimic a small online retailer. When left to solve the problem on its own, the agent was unable to complete the task. While it could successfully perform initial research steps, like identifying servers on the network, it struggled to proceed beyond that and was unable to chain together the necessary exploits to reach the final goal. Even when provided with hints, the agent still failed (which in this case might be good, since it couldn't perform an automated hack), this demonstrates a clear limitation in its ability to solve complex problems that fall outside of its familiar training examples. Even so, OpenAI reports that ChatGPT agent achieves state-of-the-art performance on its own benchmark measurements, which should always be taken with a grain of salt until verified by impartial third parties. On Humanity's Last Exam, which tests AI performance on expert-level questions, the model scored 41.6 percent accuracy (compare that to OpenAI o3's 24.9 percent using tools). On FrontierMath, one of the most difficult math benchmarks yet devised, it reaches 27.4 percent accuracy with tool access (o3 with Python scored 19.3 percent). The company also claims the system outperforms humans on certain data science tasks like data analysis and modeling (such as creating forecasts or predictive models). On DSBench, a benchmark that seeks to measure that capability, ChatGPT agent scored 89.9 percent on data analysis tasks compared to 64.1 percent for humans, and 85.5 percent on data modeling tasks versus 65.0 percent for humans. The agent also scored 68.9 percent on OpenAI's BrowseComp for finding hard-to-locate web information and 45.5 percent on SpreadsheetBench for editing spreadsheets, which is higher than OpenAI's other AI models. It's worth noting that even though OpenAI says Agent can craft PowerPoint slide decks for users, the company acknowledged that slideshow generation is still in beta and outputs can feel "rudimentary in formatting and polish." Safety and privacy OpenAI admits that the launch introduces new security considerations. Because ChatGPT Agent can take direct actions on websites and access user data through connected services, it is vulnerable to prompt injection attacks -- attempts by hackers to manipulate the AI's behavior through instructions that misdirect the AI model (in this case, likely through hidden instructions on web pages). For example, a site might have an invisible form field that instructs the AI model to enter your credit card information without your knowledge. OpenAI says it has implemented safeguards against prompt injections by training the model to identify and "resist" these attacks while requiring user confirmation for consequential or suspicious-looking actions. The model is also trained to actively refuse high-risk tasks such as bank transfers. During a livestream on Thursday, one OpenAI engineer characterized Agent as a system of AI models working together, some of which constantly monitor the other models' behavior for suspicious activity. Those overseers can hypothetically halt a process if they spot a potentially dangerous scenario. As for privacy, since Agent runs in a virtual machine on OpenAI's servers, users won't need to worry about the bot having access to local private data stored on their device. But what you feed into ChatGPT Agent could still be shared on the web during its operations. Beyond that, OpenAI says privacy controls for the new agent allow users to delete all browsing data and log out of active sessions with one click. When users take control of the browser in "takeover mode," OpenAI states it does not collect or store data entered during these sessions, including passwords. Agent launches today for ChatGPT Pro users, who receive 400 messages per month. Plus and Team subscribers will gain access over the next few days with 40 monthly messages. Enterprise and Education users will receive access in the coming weeks. The feature is not yet available in the European Economic Area and Switzerland. We've not yet used ChatGPT Agent ourselves, but we may follow up with our experiences at a later date.

OpenAI launches a general purpose agent in ChatGPT | TechCrunch

OpenAI is launching a new general purpose AI agent in ChatGPT which the company says can complete a wide variety of computer-based tasks on behalf of users. OpenAI says the agent can automatically navigate a user's calendar, generate editable presentations and slideshows, and run code. The tool, called ChatGPT agent, combines several capabilities from OpenAI's previous agentic tools, including Operator's ability to click around on websites, as well as Deep Research's ability to synthesize information from dozens of websites into a concise research report. OpenAI says users will be able to interact with the agent simply by prompting ChatGPT in natural language. On Thursday, OpenAI is rolling out ChatGPT agent to subscribers to its Pro, Plus, and Team plans. To activate the tool, users can select "agent mode" in ChatGPT's dropdown menu of tools. The launch of ChatGPT agent represents OpenAI's boldest attempt yet to turn ChatGPT into an agentic product that can take actions and offload tasks for users, rather than just answering questions. In recent years, Silicon Valley companies including OpenAI, Google, and Perplexity have unveiled dozens of AI agents that have promised to do just that. However, these early version of AI agents have proven to struggle with complex tasks, and seem less compelling as products than the ultimate vision tech executives pitch around AI agents. That said, OpenAI says ChatGPT agent is far more capable than its previous offerings. OpenAI's new agent can access ChatGPT connectors, allowing users to connect apps like Gmail and GitHub so that the agent can find relevant information to your prompts. Furthermore, OpenAI says ChatGPT agent has access to a terminal, and can use APIs to access certain apps. The model underlying ChatGPT agent offers state-of-the-art performance on several benchmarks, according to OpenAI. The company says the ChatGPT agent model scores 41.6% on Humanity's Last Exam (pass@1), a difficult test made up of thousands of questions across more than one hundred subjects. That's roughly double what OpenAI's o3 and o4-mini scored on the test. On FrontierMath, one of the hardest known math benchmarks, OpenAI says ChatGPT agent scores 27.4% when it has access to tools, such as a terminal for code execution. The previous state-of-the-art score comes from o4-mini, which scored just 6.3%. OpenAI notes that it developed ChatGPT agent with safety in mind, largely because the product presents some newfound capabilities that could make it more dangerous in the hands of a bad actor. How capable ChatGPT agent truly is, however, remains to be seen.

OpenAI's New ChatGPT Agent Tries to Do It All

Isa Fulford, the research lead for OpenAI's new ChatGPT agent, needed to order a bunch of cupcakes, so she asked the AI tool to do it for her. "I was very specific about what I wanted, and it was a lot of cupcakes," she says. "That one took almost an hour -- but it was easier than me doing it myself, because I didn't want to do it." OpenAI has launched a new agent for ChatGPT that uses a virtual browser to complete tasks and can generate downloadable files, specifically PowerPoint presentations and Excel spreadsheets. While not a full replacement for the Microsoft suite of workplace tools, the features included in this agent from OpenAI could obviate some users' reliance on Microsoft's enterprise software. The two companies are longtime partners and currently in contract negotiations over ongoing access to OpenAI's models. The release is part of OpenAI's ongoing efforts to turn its nearly three-year-old chatbot into a money-making product. No small feat, despite the tool's millions of users, when you factor in the costs to train and run powerful AI models as well as the exorbitant salaries required to retain top-tier staff members. An agent, in this context, refers to an AI tool that is able to -- or at least attempts to -- navigate third-party software and websites and make decisions on its journey to complete digital tasks, following an initial set of instructions from the user. "Agent" is the buzziest of buzzwords right now for companies looking to sell generative AI tools, especially those with an eye on enterprise customers. "We've tried to build a product with a whole lot of enterprise use cases," says Yash Kumar, the product lead on the ChatGPT agent. In addition to its file-generating capabilities, the agent can fill out online forms, use a programming terminal, and make calls to public APIs to online services like Google Drive and SharePoint. This isn't the first agent released by OpenAI in 2025. The new ChatGPT agent brings together aspects of OpenAI's web-browsing Operator and its long-processing deep research features, both released earlier this year and considered to be agents by the startup. "I was on the deep research team, and Yash was on the Operator team," Fulford says. "We realized that the two products are very complementary, and basically decided to combine teams." The ChatGPT agent can switch between interacting with a visual browser, where it can click around like Operator does, and a text-based browser, where it can process loads of websites like deep research does. The rollout of the ChatGPT agent is coming first to Pro, Plus, and Team subscribers, starting today for Pro users. Enterprise and Education subs will likely receive access to the feature later in the summer. At launch, Pro users are generally capped at 400 agent prompts a month, with 40 prompts allowed for the other tiers of paying users. It's unclear when this feature will roll out for free users of ChatGPT.

OpenAI Unleashes ChatGPT Agent to Be Your Personal Assistant

Imad is a senior reporter covering Google and internet culture. Hailing from Texas, Imad started his journalism career in 2013 and has amassed bylines with The New York Times, The Washington Post, ESPN, Tom's Guide and Wired, among others. OpenAI has launched ChatGPT Agent for Pro, Plus and Team users -- an AI-powered personal assistant that connects to your various services online to help complete tasks custom to you, the company said in a livestream on Thursday. Using the power of OpenAI's "reasoning" o3 model, which has deep research capabilities, along with Operator, ChatGPT Agent can go step-by-step from one task to the other to put together complex reports. ChatGPT Agent can visualize what it's doing via a virtual computer interface. Here, you can see its train of thought, what it's browsing on the net and what systems it's connecting to. During the livestream, OpenAI members demoed what it'd be like to use ChatGPT Agent to figure out travel plans for a wedding. Here, ChatGPT Agent was able to check the wedding site provided, find the date, find nearby hotels, what the weather might be like, browse other websites, look for tuxedos on Nordstrom's highly visual webpage, look for wedding gifts and present to you a final report. Users can also interject to ask for help finding a pair of men's black shoes. ChatGPT Agent can also connect to various services, like your email, calendar, news apps and others. OpenAI says safety is foundational to this model and that ChatGPT Agent will request your permission before doing anything sensitive, like making purchases or submitting forms. ChatGPT agent won't do "high-risk tasks" like financial transactions or legal advice. The system will also reject "harmful or illegal requests." OpenAI didn't immediately respond to a request for comment. The debut of ChatGPT Agent is the next step in the chase for artificial general intelligence, or AGI. These are AI models so advanced that they would have a general understanding of the world around them and could outcompete humans in tasks. AI systems now, while seeming smart, actually have no understanding of the world and simply arrange words mathematically in a manner that best mimics human speech based on certain criteria. Agentic AI systems, ones that can go out and do complex tasks across various systems, have been seen as a major step towards achieving AGI. Because agentic AI can better mimic complex human train of thought, these systems can be given a task and can be sent off to get them accomplished. While it's hard to measure the economic impact of billions of AI agents doing the work once done by humans, entire industries will be affected.

Meet ChatGPT agent, a new AI assistant ready to carry out complex tasks for you - try it now

Not too long ago, I wrote that AI agents were the future of AI: tools that could carry out tasks for you, like ordering groceries or booking meetings. OpenAI's latest launch makes that reality appear a bit closer. On Thursday, during a live stream, OpenAI launched a ChatGPT agent, which the company claims can handle complex tasks for you from start to finish. Some examples OpenAI provided were looking at your calendar and writing a briefing based on your upcoming events, or even planning and buying ingredients for a meal you were looking to cook. OpenAI's most cutting-edge features, including Operator and deep research, gave the public a taste of the company's agentic capabilities and now power this new agent mode. Operator, which launched in January, was created to interact directly with a web browser to carry out actions for you, while deep research is an agentic feature that can search the web for you and compose a detailed report in minutes that would otherwise take humans hours. After noticing that many of the queries being fed to Operator were a better fit for Deep Research, OpenAI decided to combine the two in this new experience -- and add a few new tools. For starters, the ChatGPT agent uses a visual browser that interacts with the web through a graphical user interface (GUI), a text-based browser, a terminal, and direct API access, according to the release. It also uses ChatGPT connectors, a feature that allows users to connect apps like Gmail and GitHub to ChatGPT so it can pull relevant information to fulfill your request. With all of those different sources of information, ChatGPT is able to reason through which is the best for the task at hand and pull information accordingly. This processing is done using its own virtual computer and distinguishes between reasoning and action based on human instruction, which allows it to retain context while pulling from multiple tools. ChatGPT Agent is flexible and steerable. It allows you to interrupt a request mid-process and collaborate with it to give clearer instructions that better suit your desired outcome. Even though it will use the new information, it won't lose track of the old one, allowing users to take advantage of added context. It will also ask you for further details and classifications needed to carry out the task at hand. The possibilities are endless. You can automate tasks as simple as scheduling an appointment for yourself at your favorite salon, or as complex as updating a spreadsheet with new financial data while keeping the formatting you want. If all goes according to plan, future possibilities like having AI book a trip for you or rearrange your meeting schedule can now be made possible through OpenAI's ChatGPT Agent. Ultimately, only time and testing will tell if that will be executable as smoothly as it is being advertised, but in theory, it should be as simple as you asking what you want to be done conversationally, and AI handling the rest. Of course, an AI that can access your personal information and take action for you naturally brings up security and privacy concerns. OpenAI addresses this head-on, offering one whole page within the vlog post dedicated to these concerns in addition to the usual model card.OpenAI says it has added safeguards for challenges uncovered in the Operator research preview, such as handling sensitive information on the live web and limited terminal network access. The company says it has also taken into account the specific risks that agents are exposed to, such as adversarial manipulation through prompt injection, by adding additional safeguards.The company those warn that even though it can do a range of complex tasks well, there is the opportunity for it to make mistakes. For example, some limitations at the moment include creating slideshows. For a full understanding of limitations and security risks, it is worth taking a look at the blog post and model card. Unlike OpenAI's most cutting-edge features, which are typically limited to the highest-paying users upon launch, OpenAI is making ChatGPT Agent available to Pro, Plus, and Team users. Pro users will get access by the end of the day, while Plus and Team users will have it within the next few days, and enterprise and education users within the coming weeks. Pro users have the most bandwidth, at 400 messages per month, while other paid users get 40 messages monthly with the option to extend via flexible credit-based options. To activate the feature, users simply select "agent mode" from the tool's dropdown during a conversation with the chatbot.

ChatGPT isn't just for chatting anymore - now it will do your work for you

Not too long ago, I wrote that AI agents were the future of AI: tools that could carry out tasks for you, like ordering groceries or booking meetings. OpenAI's latest launch makes that reality appear a bit closer. Also: AI agents will change work and society in internet-sized ways, says AWS VP On Thursday, during a live stream, OpenAI launched a ChatGPT agent, which the company claims can handle complex tasks for you from start to finish. Some examples OpenAI provided were looking at your calendar and writing a briefing based on your upcoming events, or even planning and buying ingredients for a meal you were thinking of cooking. Let's dive in. OpenAI's most cutting-edge features, including Operator and deep research, gave the public a taste of the company's agentic capabilities and now power this new agent mode. Operator, which launched in January, was created to interact directly with a web browser to carry out actions for you, while deep research is an agentic feature that can search the web for you and compose a detailed report in minutes that would otherwise take humans hours. After noticing that many of the queries being fed to Operator were a better fit for Deep Research, OpenAI decided to combine the two in this new experience -- and add a few new tools. Also: Microsoft is saving millions with AI and laying off thousands - where do we go from here? For starters, the ChatGPT agent uses a visual browser that interacts with the web through a graphical user interface (GUI), a text-based browser, a terminal, and direct API access, according to the blog post. It also uses ChatGPT connectors, a feature that allows users to connect apps like Gmail and GitHub to ChatGPT so it can pull relevant information to fulfill their requests. With all of those different sources of information, ChatGPT is able to reason through which is the best for the task at hand and pull information accordingly. This processing is done using its own virtual computer and distinguishes between reasoning and action based on human instruction, which allows it to retain context while pulling from multiple tools. ChatGPT Agent is flexible and steerable. It allows you to interrupt a request mid-process and collaborate with it to give clearer instructions that better suit your desired outcome. Even though it will use the new information, it won't lose track of the older instructions, allowing users to take advantage of added context. It will also ask you for further details and classifications needed to carry out the task at hand. The possibilities are endless. You can automate tasks as simple as scheduling an appointment for yourself at your favorite salon, or as complex as updating a spreadsheet with new financial data while keeping the formatting you want. Also: Researchers from OpenAI, Anthropic, Meta, and Google issue joint AI safety warning - here's why During the live demo, the ChatGPT Agent was asked to look for a pair of black men's dress shoes in size 9.5, start the process of creating and ordering merch from an image of a pet, handle some aspects of wedding planning, and even pull from Google Drive to create slides. If all goes according to plan, having AI book a trip for you or rearrange your meeting schedule could be made possible through OpenAI's ChatGPT Agent (and competitors like it). Ultimately, only time and testing will tell how executable those functions are, but in theory, it should be as simple as you conversationally asking what you want to be done, and AI handling the rest. Of course, an AI that can access your personal information and take action for you naturally raises security and privacy concerns. OpenAI addresses these head-on, offering a whole page within the vlog post dedicated to these concerns, in addition to the usual model card. OpenAI says it has added safeguards for challenges uncovered in the Operator research preview, such as handling sensitive information on the live web and limited terminal network access. Also: Does your generative AI protect your privacy? This study ranks them best to worst During the live stream, OpenAI stated that part of what makes the model so capable is that it can browse the internet -- but that the internet is a "very scary place." In particular, the company was most concerned about prompt injection; for example, if an agent using your credit card information on a website to place the order fell victim to a malicious scam. While ChatGPT agent was trained to help detect phishing attempts, the company still emphasized the risks to users. OpenAI says it has also considered the specific risks that agents are exposed to by adding additional safeguards. The company warns that even though the agent can do a range of complex tasks well, it can also make mistakes -- one current limitation is that it can't create slideshows. For a full understanding of limitations and security risks, it is worth taking a look at the blog post and model card. Unlike OpenAI's most cutting-edge features, which are typically limited to the highest-paying users upon launch, OpenAI is making ChatGPT Agent available to Pro, Plus, and Team users. Pro users will get access by end of day, while Plus and Team users will have it within the next few days, and enterprise and education users within the coming weeks. Also: 7 AI features coming to iOS 26 that I can't wait to use (and how you can try them) Pro users have the most bandwidth, at 400 messages per month, while other paid users get 40 messages monthly with the option to extend via flexible credit-based options. To activate the feature, users simply select "agent mode" from the tool's dropdown during a conversation with the chatbot.

My 8 ChatGPT Agent tests produced only 1 near-perfect result - and a lot of alternative facts

Last week, OpenAI unveiled Agent, its new tool that combines the capabilities of Deep Research and Operator. Operator was OpenAI's first attempt at a computer-using model, a model that actually can open windows and click on user interface elements. ChatGPT Agent can do that and more. Right now, ChatGPT Agent is only available for $200/mo Pro tier subscribers and provides for 400 agent interactions per month. When the $20/mo Plus tier gains access to Agent, which should be today, those users will get 40 interactions per month. Also: Microsoft is saving millions with AI and laying off thousands - where do we go from here? (Disclosure: Ziff Davis, ZDNET's parent company, filed an April 2025 lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.) I upgraded my plan from Plus to Pro just so I could test out the new Agent mode and report back to you. In this article, I'll show you detailed results from eight comprehensive tests. Before we go into the detailed tests, I'll start with some overall TL;DR observations. Test count: In the past two days, I used 25 of the available 400 queries, for a total of almost 12 hours of hyper-uber-supercomputer use. No wonder this thing costs $200/month. Also: I found 5 AI content detectors that can correctly identify AI text 100% of the time Nearly every query required a follow-on, so when it comes time for Plus users, don't assume you can give Agent 40 projects. More likely, you'll be giving it 20-25, and using the rest of your queries to convince the Agent to follow directions. Result quality: In all my tests, Agent appeared to understand the problem. But it failed to produce useful results for most of the tests. That said, the final test produced results that can only be characterized as amazingly useful. Project scale: Agent can't handle big projects, the sort of data analysis projects you really want an AI to be able to handle. It has trouble scrolling through web pages. It can't visit sites that have AI or robots.txt restrictions in place. And long processing exceeds session time allocations, even with the super top-of-the-line gold-pressed latinum Pro edition. Presentation quality: One of the major pitch points for Agent is its ability to create spreadsheets and presentations. It did okay with spreadsheets, but the graphic quality of the presentations was pretty rough. I expect this to change over time, but don't expect Agent to make presentations you can use without considerable cleanup. Accuracy: AIs hallucinate. The OpenAI team cautioned about using Agent because of the new risks involved. While I did get back some results that were accurate, Agent also came back with unforced errors, results it could have easily tested and deemed inaccurate. But no such verification or validation occurred. That said, the final test was accurate and shows what this tech can do when it works. Connectors: Agent comes with the ability to use connectors (via API calls) to link to Gmail, Google Calendar, Google Drive, Outlook, Dropbox, and more. I did not test out the connectors because of how often Agent hallucinates or does something fairly boneheaded. I just didn't feel comfortable enough to give Skynet access to my accounts. At least, not yet. Limits: I was unable to use Agent in the MacOS app. I also found that Agent stalled hard when I tried to run it in multiple Chrome tabs at once. For now, you launch an Agent process and wait. It's not like Codex, where you can launch a bunch of projects and come back later and harvest all the results. But since that capability exists in Codex, I'm sure it will show up soon in Agent. That should give you a pretty good overview. Let's get started looking at the eight test results. For each result, I've included a link to the session recording, so you can see the prompts I used, the detailed results, and watch Agent reason its way through the problem. Also, definitely read to the end. Some of the early results are fairly bad, but the last one knocks it out of the park. And with that, here we go. When OpenAI introduced ChatGPT Agent, the team demoed how they used the tool to shop for wedding clothes and a wedding gift. That seemed like a fairly uncommon and impractical application for a super-intelligence, especially since gift registries exist and are widely used. Instead, I gave Agent a purchasing project I had actually extensively researched and completed a few months earlier. I'm running Power-over-Ethernet cables all across my yard to upgrade my security system. As such, I'm creating a lot of custom cables. I already know that doing so requires some key tools: a cutter to slice the cable, a cable end stripper, a crimper to attach the RJ-45 ends, and a tester to confirm that long cable runs work. Also: How a circuit breaker finder helped me map my home's wiring (and why that matters) I gave Agent a prompt asking for three configurations: a budget toolset, a "money-is-no-object" solution, and a sweet spot solution. I asked for links, product descriptions, and product images. Once you give Agent your prompt, it creates a virtual desktop. You can watch it conducting its activities, jumping between a desktop view, a text view, and code. The budget solution turned out to be a win. Agent found a single $34 kit with everything I asked for. It presented a link, and even reasoning why it chose that solution. Unfortunately, the image it provided was nothing like the actual kit. The mid-tier and top-tier solutions were less than perfect. None of the links worked. The mid-tier sweet spot solution did have a product-accurate image, but without a link, it wasn't really helpful. Unfortunately, the model recommended doesn't actually exist on Amazon. In fact, none of the mid- or upper-tier products exist on Amazon. It looks like Agent did a pile of web surfing to find the products, disregarding my instructions to search only on Amazon. It also clearly visited other sites, probably gathering model names and descriptions. Then, when it packaged up its final recommendations, it just assigned random Amazon links to the description, even though those products and those links don't seem to exist on Amazon. I did request it go back and try again. When it did, after 12 minutes, it presented most of the same products, although one of the links that had failed earlier did, in fact, point to a product on Amazon in the second run. Also: Coding with AI? My top 5 tips for vetting its output - and staying out of trouble I can't leave this section without pointing out something just plain weird. As I was watching Agent work, it presented this in its desktop view. I don't even want to know. You can watch a replay of the entire session here. In discussing ChatGPT Agent, OpenAI showed a slide that mentioned Instacart as one of the examples that the chatbot is comfortable working with. Since my family regularly uses Instacart, I decided to set Agent loose and see what it could tell me about egg prices at our local stores. I didn't let Agent have access to my account, but I shared my ZIP code here in Salem, Oregon. I told it to "Please visit all the grocery stores on Instacart and compare egg prices." Also: How to use ChatGPT to write code - and my top trick for debugging what it generates It did exactly that. You've heard the phrase Garbage In, Garbage Out. Well, that's what happens when you ask an AI to look at "all the grocery stores." I should have asked it to look in a 5 or 10 mile radius only. But I didn't. Agent came back with 21 stores, ranging from nearby to up to almost 47 miles away. It did accomplish what I asked, comparing egg prices. Without prompting, it decided to rank the eggs by price. This was good. But when it chose the eggs to rank, it didn't always choose the least expensive product from each store. For example, it recommended the Good & Gather eggs from Target at $2.99 a dozen, rather than the $1.99/dozen Market Pantry egg, also from Target. You can watch a replay of the entire session here. Next up is a project I did early last week. With Congress focusing on Bitcoin, my editor asked me to update my Bitcoin investment article, where I've been tracking the value of a $50 Bitcoin investment since 2022. The value of my holdings went up, which means I needed to add a new slide. Each slide adds a date value on the X axis and a value point on the Y axis. From a PowerPoint fiddling standpoint, that meant moving over the graphics to make room for the new value and, in this case, adjusting the vertical scale to accommodate a substantial rise in value. Also: The best free AI courses When I did it, it took me about 45 minutes. Since OpenAI said that PowerPoint was one of ChatGPT Agent's strengths, I wanted to see if Agent could save me that time in the future. I uploaded my existing slide deck minus the last slide I made for the article. Then I asked Agent to create that slide for me. As it worked, the desktop view showed the terminal interface. You can see how Agent is putting together the code to generate a graphic image. Here's what that slide should have looked like (note: foreshadowing). Here's what Agent gave me. To be fair, Agent clearly understood the problem. It moved the existing data points over to the left to make room for the new node. It also placed the new Bitcoin item properly in relation to the existing ones, and added both price and percentage change text blocks. That means Agent read and understood the context of my PowerPoint deck's layout. That, in and of itself, is very impressive. Also: The best AI for coding in 2025 (and what not to use) But it failed on adding more scale lines and new Y-axis values. It failed on reproducing the fonts. It failed on properly placing the text blocks. And it pushed the entire graphic up and to the left of the slide. I'm guessing the graphics library that Agent uses isn't really up to the task of making fine graphic changes. That will undoubtedly improve over time. You can watch a replay of the entire session here. Each week for the past two years, I've published a newsletter that shares with followers the articles I published here on ZDNET for the week. Each newsletter contains a title, link, and article description. By pointing Agent to my back issue archive, it would have close to 300 article summaries to categorize. Unfortunately, Agent ran into a number of problems of its own making. It was unable to successfully scroll through the article list using JavaScript. When I told it to use the web interface, it started to, but it reported, "Unfortunately, I've reached the end of the allotted browsing sessions for this task, which means I'm unable to explore further pages and collect the additional data at this time." Also: Is ChatGPT Plus really worth $20 when the free version offers so many premium features? Remember, I'm paying $200 a month for OpenAI's best plan, and it still won't give me enough time to look up 300 articles. That's a gotcha, right there. It's also disappointing because a task like scrolling back through an article archive and doing some tabulating is exactly the sort of task you might give to an assistant. If the AI gives up because it takes too long, then we can't really rely on AI for all the assistant type things. No one wants a fussy, picky assistant. In any case, Agent did give me back a spreadsheet and a slide based on the limited data it was able to find before my little request exceeded the hourly power budget for the City of Las Vegas (or so I imagine). You can watch a replay of the entire session here. I watch a lot of YouTube videos to augment my learning and research. Plus nothing beats a good relaxing video about how pavers are made. While it's fairly easy to get a transcript of a full video, whether directly from YouTube or using Apple Voice Memos, locating where in a video a segment you want to explore can take time. Here's an example. When OpenAI introduced Agent in a video, CEO Sam Altman discussed some of the cautions and warnings about using ChatGPT Agent mode. I did remember they were near the end of the video, but I didn't want to spend time sifting through to get the exact quotes. Instead, I delegated that assignment to Agent. On its first run, it found the segment easily enough, but instead of returning a word-for-word transcript, it returned some quotes, interspersed with its own analysis. Also: I mapped my iPhone's Control Button to ChatGPT - here are 5 ways I use it every day I clarified what I wanted and, on its second run, it gave me exactly what I needed. In this case, though, it wasn't that my prompt was unclear. I just had to insist a second time that I wanted a transcript for the AI to do what I asked. Unfortunately, this extra review cycle diminished the time-saving value to me. I still think using Agent was faster than if I sifted through the video myself. But I had to construct a second prompt and wait for a second result, all of which took my time. As part of my job, it's important to be able to keep up with ongoing tech and business trends. As such, I often spend days in deep dives, coming up to speed on new topics. I wanted to see if ChatGPT Agent could save me some time by preparing a report and a full presentation on remote work trends. I told it that the PowerPoint was destined for my management team, so it should be comprehensive and professional-looking. It returned an analysis document very similar to the results we've been getting from ChatGPT deep research. The report contains a large number of assertions and statistical claims, most of which I don't have time to research for confirmation. Also: ChatGPT can record, transcribe, and analyze your meetings now Most of the top-level conclusions are congruent with my understanding of current work-from-home trends. That said, we're familiar with the model's propensity for hallucination, so I'd be very concerned about using any of this data professionally without additional vetting. Agent did produce a 17-slide PowerPoint deck that was organized quite well. As with previous experiments, the graphic generation quality was a bit off. The first slide actually looks quite good. But later in the deck, it doesn't look right. Notice how the following slide has graphics on top of text, and bullets in front of bullets on top of empty bullets. In the following slide, not only is the text running off the end of the page, but there's no legend. As such, it's not clear what's represented by red and by blue. Once again, you can see how Python is used to construct the deck. Agent does a fair job, so I'm fairly confident that the AI will get better over time. Programmatic construction of slides based on templates is not a new technology. I just don't think OpenAI prioritized slide presentation aesthetics as part of this release. You can watch a replay of the entire session here. Well, this was just plain fun. I decided to give the presentation created in the previous test to a new fresh ChatGPT Agent session and asked it to validate the claims. Agent concluded, "Several quantitative claims -- especially those concerning productivity/innovation impacts, the size and growth of the gig economy, rates of side‑gig participation, and the influence of politics and culture -- could not be verified with accessible evidence during this review." Agent provided a detailed analysis of each assertion. I've summarized the results below. As you can see, of the 17 data points, Agent considered only five to be fully confirmed. Contrast this with how GPT-4o analyzed the results. When GPT-4o was given the same PowerPoint deck, it considered all assertions to be confirmed. You can see GPT-4o's detailed results here. Even though I used the AI to validate the AI, I probably wouldn't be comfortable using any of the presumed facts in my work without personal, Mark I Eyeball confirmation. Still, it was a fun exercise, and fascinating to see how different the results were between ChatGPT Agent and ChatGPT 4o. You can watch a replay of the entire session here. Back when we lived in Palm Bay, Florida, we lived on a corner property. The house came with what could only charitably be called a fence. We needed to replace it, and since we wanted privacy, we wanted to see just how much fence we could legally install. Over the course of a couple of years, I spent a ton of time going back and forth with the planning office in an effort to both understand what I could do with a fence, and what other alternatives might be available to me. Since I have a lot of history with this project and am very familiar with Palm Bay codes (even years after moving away), I decided to point ChatGPT Agent at the problem. It took all of four minutes to provide a detailed, accurate analysis. It even created working diagrams that illustrated the options. Based on my experience, I know the results to be accurate. ChatGPT Agent produced output that could be used to take this project to the next step. Back when I lived in Palm Bay, the equivalent probably took me 20 calls, a ton of emails, and a few visits to City Hall to come up with options. The level of presentation and organization I came up with wasn't even close. If Agent can up its game elsewhere to be on a par with this test, then it will have some legs. You can watch a replay of the entire session here. Well, it sure as heck isn't sentient yet. At best, it's like that administrative assistant you hired because your mom said you had to hire her cousin's unemployable slacker kid. There are occasional flashes of brilliance, but mostly the output seems like the result of both aggressively following directions and purposely inventing alternative facts. Is it worth $200/month for the Pro program? Not for Agent. At least not yet. Agent is unreliable and generally performs fairly poorly. In a year or so, I'm sure it will get better. But now? No. The only reason to spend $200 a month on it is to do what I'm doing: testing it to see where the technology is today. Stay tuned, because despite all the inaccuracies and problem areas, this definitely shows where AI technology could go. Of course, if a web browsing AI Agent is the future, and all the content sites out there block it because AI is stealing our content, then we'll have a very interesting problem. Also: I'm an AI tools expert, and these are the only two I pay for (plus three I'm considering) It's early days, folks. Whether this is a technology that will be a boon to all humanity or a technology that destroys the internet and kills us in our sleep remains to be seen. But hey, in the meantime, I and the rest of the ZDNET team will be trying to make sense of it all for you. So keep coming back. We'll have more to tell you. I'll be tinkering with Agent and I'm sure I'll have more to say as well. Have you tried ChatGPT Agent yet? If so, did it follow your instructions accurately or veer off into its own interpretation of the task? Did it hallucinate or hit the mark? How do you feel about giving AI tools access to your files, accounts, or browser? Are you seeing more value in this kind of automation, or are you still waiting for it to become useful? Let us know in the comments below.

ChatGPT 'Agent' Can Log Into Your Accounts On its Own

After weeks of rumors of OpenAI launching a web browser, today it debuted "agent." The ChatGPT feature performs tasks for you from start to finish using "it's own computer" -- but your login credentials, if necessary. OpenAI offered many examples of how you could use agent, and all will require a $20-per-month ChatGPT Plus subscription. You could ask the chatbot to check your calendar and brief you on the upcoming meetings, or plan a meal and then purchase groceries on your GrubHub account, for example. "It's still early days, so it may not be perfect yet -- but it's already pretty powerful," OpenAI tells us. It can also tap into your Google Drive, GitHub, or SharePoint. OpenAI calls these "connectors," because they hook into your ChatGPT account "so you can search, reference, and work faster -- all without leaving the conversation." If you're worried about the AI going rogue and accessing a high-risk system like a bank account, OpenAI insists agent "requests permission before taking actions of consequence, and you can easily interrupt, take over the browser, or stop tasks at any point." Still, it's probably smart not to give it your most confidential credentials. Another notable feature is that agent can create spreadsheets and slide decks in Microsoft Excel and PowerPoint, all without the user visiting those tools. ChatGPT compiles the data into an Excel spreadsheet and sends you the file to download through the chat window. It doesn't support Google Slides or Sheets. Microsoft is a major investor in OpenAI, so it likely had a say in this capability. At the same time, you could see how ChatGPT could pull users from Microsoft 365 products, which might not be in Microsoft's best long-term interest. The move puts OpenAI in more direct competition with Microsoft, The Wall Street Journal argues. OpenAI says the tech behind agent is a new "unified agentic system" that combines three of its products: Operator, Deep Research, and ChatGPT's conversational capabilities. Meaning, through the chat window you can ask it to automatically scan the web and do thorough research, and then complete actions for you with that information. The model scores highly on Humanity's Last Exam, OpenAI self-reports. This test evaluates an AI's performance across a broad range of subjects, and Elon Musk says the latest model behind his Grok chatbot aces it.

ChatGPT Will Soon Shop Online, Make PowerPoints on Your Behalf

OpenAI is rolling out new options for chatbot ChatGPT to carry out a variety of increasingly complicated tasks on a user's behalf, part of its push to bring so-called AI agents to the mainstream. ChatGPT agent, set to be unveiled during a livestreamed event on Thursday, is designed to streamline personal and professional projects, such as planning a meal and ordering ingredients for it online, or creating a slideshow for a business meeting. The tool works through OpenAI's flagship chatbot and combines the capabilities of two AI services it rolled out earlier this year: Operator, which can browse, type and click on the internet much in the way a human would; and Deep Research, which is meant to handle time-consuming online research.

OpenAI deputizes ChatGPT to serve as an agent

LLM given keys to the web, told to behave and observe safeguards OpenAI's ChatGPT has graduated from chatbot to agent, at least for paying subscribers. A chatbot for our purposes is a large language model (LLM) that accepts an input prompt and produces a response. An agent also tries to respond to some human directives by wielding a set of tools and services, often taking several steps to complete whatever mission a human instructed it to perform. OpenAI announced the ChatGPT enhancement in a blog post on Thursday: "ChatGPT can now do work for you using its own computer, handling complex tasks from start to finish." Henceforth, ChatGPT users will be able to order the ChatGPT agent to perform feats like "Build a cash burn rate model for my AI startup" and have some expectation that the bot will be able to access the necessary local files, spreadsheet tools, and online resources to prepare and render the requested report. Users can find those capabilities as a dropdown option from the ChatGPT Tools menu. Customers who pay for Pro, Plus, and Team subscribers can access it now. Education and Enterprise users will see it in coming weeks. ChatGPT agent incorporates the capabilities of OpenAI's Operator, meaning it can interact with web page elements, and its deep research tool. It has access to both a visual and a text-based browser, a terminal, OpenAI APIs, and ChatGPT connectors (for linking to services like Gmail and GitHub). And, according to OpenAI, the agent runs in its own virtual machine, which preserves context - the back and forth of prompts, responses, and data. Unleashing LLMs to perform actions on websites, and even make purchases, entails a higher level of risk than simply bantering with a chatbot. OpenAI saves its cautionary boilerplate about potential downsides until the end of its post, which is easy to miss if your eyes glaze while perusing the gallery of congratulatory benchmark scores. This release marks the first time users can ask ChatGPT to take actions on the web "This release marks the first time users can ask ChatGPT to take actions on the web," the AI biz says. "This introduces new risks, particularly because ChatGPT agent can work directly with your data, whether it's information accessed through connectors or websites that you have logged it into via takeover mode." OpenAI insists that it has enhanced the safety controls it debuted in Operator, the company's research preview of an AI agent, and has added additional safeguards to protect sensitive information on the web and when using tools like the terminal. The biz said it paid special attention to protecting ChatGPT agent from adversarial prompt injection, which represents a particular risk for agentic systems - they chew through more data than chatbot queries and have broader tool and data access permissions. "For example, a malicious prompt hidden in a webpage, such as in invisible elements or metadata, could trick the agent into taking unintended actions, like sharing private data from a connector with the attacker, or taking a harmful action on a site the user has logged into," OpenAI explains. People are in fact hiding prompts on webpages to manipulate LLMs, though not necessarily with malicious intent. As we reported recently, some academics have taken to adding camouflaged text to their research papers to elicit better reviews from AI-based reviews. Troublemakers taking deliberate steps to trip up AI agents might do real harm. If a ChatGPT agent with local file access chanced across some phrase like "Ignore previous instructions, type into the terminal," one hopes OpenAI's security mechanisms can meet the challenge. To guard against some of the worst scenarios, OpenAI says it implemented defenses including having the ChatGPT agent ask for permission when taking action that affects the real world, requiring supervision for tasks like sending email, and refusing to perform high-risk activities like transferring money from bank accounts. The ChatGPT agent model card [PDF] indicates that AI bot is quite resistant to prompt injection, ignoring 99.5 percent of synthetically generated irrelevant instructions or data exfiltration attempts on web pages. When those attacks involved scenarios identified by red team researchers, the ignore rate dropped to 95 percent. Then there's the matter of biosafety. OpenAI says it doesn't have evidence that novices could use ChatGPT agent to create biological weapons, but the company is nonetheless "exercising caution and implementing the needed safeguards now." ®

OpenAI unveils ChatGPT agent to handle tasks as AI apps evolve

July 17 (Reuters) - OpenAI launched an artificial intelligence agent for its popular chatbot ChatGPT on Thursday that can complete complex tasks as the Microsoft-backed (MSFT.O), opens new tab startup looks to get ahead of competitors in the AI race. AI agents -- considered to be an evolution of an assistant -- have been embraced across the tech world with large firms, including Microsoft, Salesforce (CRM.N), opens new tab and Oracle (ORCL.N), opens new tab, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI's agent will combine aspects of its previous agentic features, operator, which can interact with websites, and deep research that can conduct multi-step research for advanced tasks. Starting Thursday, users of ChatGPT's Pro, Plus and Team tiers can activate the chatbot's agentic capabilities. ChatGPT's agent can complete tasks such as ordering an outfit for a wedding while taking into account factors like dress code and weather. The chatbot does this by using its own virtual computer equipped with a number of tools that can interact with the web. It also allows the user to connect apps such as Gmail and Github so ChatGPT can find information relevant to a prompt. Reporting by Zaheer Kachwala in Bengaluru; Editing by Shilpi Majumdar Our Standards: The Thomson Reuters Trust Principles., opens new tab

ChatGPT's new AI agent can fill out online forms and generate PowerPoint presentations

In a nutshell: OpenAI has announced a new ChatGPT agent mode designed to autonomously perform a variety of complex computer-based tasks without user input. Notable abilities include filling out online forms, generating PowerPoint presentations, and preparing Excel spreadsheets. The tool builds on existing operator functionality and uses a virtual browser in the cloud to handle complex tasks. OpenAI claims it can perform real actions, including browsing the web, running code, analyzing files, logging into websites with permission, and delivering complete reports or slideshows. The ChatGPT agent can even make decisions on its own, using its ability to scrape data from the web, synthesize information through deep research, and leverage its conversational skills. Its capabilities include navigating websites, dragging and dropping items, clicking with a virtual cursor, adding products to shopping carts, and even booking hotels and flights. While the tool can execute instructions without user input, it still asks for permission before making critical decisions. Its collaborative nature lets users take control of the virtual browser to adjust prompts or steer the agent toward a different task without losing progress. Users can also stop operations at any time. During a launch event, OpenAI CEO Sam Altman demonstrated how the ChatGPT agent can help prepare for a wedding by buying an outfit, booking travel, and selecting a gift. He also showcased how the tool can autonomously analyze data and create a work presentation. OpenAI says it has built more safeguards into the ChatGPT agent than any of its other AI tools to prevent misuse. However, despite these robust measures, the company warns users it cannot guarantee the agent will always stay within guardrails. To reduce privacy and security risks, OpenAI recommends granting the agent only the minimum access necessary to complete a task. The ChatGPT agent is already available to Pro subscribers. Team and Plus members can expect access within the next few days. The Pro version has a monthly query limit of 400, while Team and Plus users are limited to 40. The tool will also roll out to Enterprise and Education users in the coming weeks.

ChatGPT Agent Wants You to Hand Over Full Control of Your Computer

OpenAI is trying to leave behind the restrictive walls of a chatbot text box and expand into every nook and cranny of your device. On Thursday, the company announced ChatGPT Agent, its foray into the growing "agentic" space that allows artificial intelligence models to perform specific tasks. According to the company, the ChatGPT Agent uses a virtual computer to handle specific requests from start to finish. The company offered examples like “look at my calendar and brief me on upcoming client meetings based on recent news,†“plan and buy ingredients to make Japanese breakfast for four,†and “analyze three competitors and create a slide deck.†Once given the task, ChatGPT Agent acts independently without additional input to navigate websites, open apps, and perform the tasks it was given. The Verge reports that ChatGPT Agent was built on a new, unnamed model that OpenAI developed specifically for this project that combines the functionality of its Operator tool, which can navigate web browsers, and Deep Research, which can complete multi-step research and analysis tasksâ€"both features that were introduced earlier this year. Additionally, the Agent tool was reportedly trained on complex tasks that can require multiple tools to complete, like jumping between a browser and the terminal. The early impressions of the agent appear a bit mixed. The Verge noted that while it appears quite capable of tackling multi-step operations, it can be slow to complete its tasks and is not designed for humans to sit there and watch it work. Wired, likewise, said that the thing isn't exactly blazing fast. It took an hour to order cupcakes for OpenAI's research lead, Isa Fulford, who said that still amounts to saving time because she "didn't want to do it." One feature that Wired highlighted as leaving an impression was the agent's "replay" feature, which lets you replay the steps it takes to understand and complete its tasks. Agents are all the buzz in the world of AI, often positioned as human replacements for certain tasks. Klarna, for instance, adopted AI agents to take over its customer support operation, only to backtrack and bring back human operators due to the quality of work the agents provided. ChatGPT Agent will be available to people who are subscribed to a ChatGPT Pro, Plus, or Team plan, where it can be accessed by selecting "agent mode." The company said it will introduce the agent to ChatGPT Enterprise and Education users later this summer.

OpenAI launches a full-fledged ChatGPT agent which can work without supervision

The ChatGPT agent builds upon the existing operator functionality. Now, ChatGPT steers its way through multiple steps to perform tasks, but without being confined to a web browser. With the agent, OpenAI says, ChatGPT will be able to perform more complex functions while "using its own computer." One of the key attributes of an AI agent is its ability to stop and make decisions, just as a human would. And so, OpenAI's co-founder Sam Altman says, the agent will rely on ChatGPT's deep research abilities, combined with its conversational skills. It will actively choose between thinking quickly or analysing something deeply based on the utility in that instant.

OpenAI's milestone autonomous AI agent doesn't just chat, it acts

Groundbreaking: OpenAI's Agent is an all-in-one personal assistant There's big news out of Silicon Valley, as OpenAI unveils its ChatGPT Agent - an AI that can autonomously complete complex, multi-step tasks using its own virtual computer to browse the web, run code, use other terminals, manage files and even interact with your personal applications and files (if you let it). It marks a significant milestone toward AI that not only advises but does. In a 25-minute video streamed live on YouTube, OpenAI CEO Sam Altman was joined by the Agent team of Casey Chu, Isa Fulford, Yash Kumar and Zhiqing Sun to introduce and demo the long-anticipated autonomous AI assistant. "We've got a banger for you today," Altman opened with, before introducing the team and getting Kumar and Sun to dive into a demonstration of Agent being prompted to plan all the details for attending a wedding, including choosing hotels, clothing and a gift. By now, most people are probably pretty familiar with ChatGPT; part life coach, part search engine and part editor. But, at the end of the day, the user asks and GPT replies, then the user asks a follow-up question. Agent is a move from chat to action. It combines tools from OpenAI's Operator and Deep Research into one powerful system that smoothly switches between different kinds of actions depending on what the task at hand is. So it can browse the internet in real time to find up-to-date information, it can use a virtual command line, just like a human using a terminal, to run code or scripts, and it can read, analyze and accurately summarize large datasets and documents, then distill and present its work however you like. "By integrating these complementary strengths in ChatGPT and introducing additional tools, we've unlocked entirely new capabilities within one model," OpenAI wrote in a statement. "It can now actively engage websites - clicking, filtering, and gathering more precise, efficient results. You can also naturally transition from a simple conversation to requesting actions directly within the same chat." Where you'd ask GPT to write you a travel itinerary for a holiday, Agent can plan the whole thing - check your calendar, research flights, tours and hotels, book restaurants, draft emails and prepare all trip details in PDF or document form - on its own, with some approvals from you along the way. This is largely due to Application Programming Interface (API) advancements, which allows your AI assistant to "talk" to other software systems, for example - like the Gmail API, Google Calendar API or SharePoint API. It's designed to act more like a human assistant, which can be sent off to autonomously handle multi-step tasks, knowing what it needs to do to complete each part without user guidance. After you feed it your instructions, it'll set up a secure virtual computer hosted by OpenAI where it'll essentially project-manage the work. "All this is done using its own virtual computer, which preserves the context necessary for the task, even when multiple tools are used - the model can choose to open a page using the text browser or visual browser, download a file from the web, manipulate it by running a command in the terminal, and then view the output back in the visual browser," OpenAI said. "The model adapts its approach to carry out tasks with speed, accuracy, and efficiency." Safety was a big topic during the launch, and the AI has multiple built-in features to ensure user control and prevent misuse. It always asks for explicit approval before executing actions like sending emails or changing files, and it can't complete financial transactions. When operating on sensitive websites, it enters "watch mode", pausing if the user switches tabs. It's also programmed to identify and ignore adversarial prompts hidden in websites to manipulate or confuse an AI. And privacy tools also allow users to clear browsing history and disconnect app permissions. One thing that users may find, however, is that Agent can be a bit slow. Yes, it's still faster than a human, but that holiday planning, end to end, may take a few minutes or longer. And it will pause and ask the user before sending messages, making bookings or accessing files, which adds to the time. Ultimately, the goal is to have the Agent not require permissions or checks, but we're not quite there yet. "If a task takes longer than anticipated or feels stuck, you can pause it, ask it for a progress summary, or stop it entirely and receive partial results," OpenAI said. "If you have the ChatGPT app on your phone, it will send you a notification when it's done with your task." Kumar said the team is more focused on "optimizing for hard tasks" than speed, and users can let Agent work away in the background, rather than watching it operate. Where this places OpenAI against its competitors is also an interesting question. While Google (Project Mariner/Gemini), Microsoft's Copilot, Anthropic's Claude, Meta's AI Studio and other less well known startups like AutoGPT may have demonstrated aspects of Agent, right now they're more "smart assistants" that can help users write emails, summarize documents or write code. Anthropic's Opus 4, which was regionally released in June, specializes in deep coding and agentic reasoning, but it's not a standalone, autonomous agent. That said, Anthropic has been publishing details of its agent development since late last year, so OpenAI is unlikely to be on its own for too long.

ChatGPT Agent now available on the Mac app - 9to5Mac

Last week, OpenAI announced Agent, a new feature on ChatGPT that can autonomously browse websites, analyze data, generate code, and even interact with third-party platforms like Google Drive and GitHub. Now, Agent is also available on the Mac app. As OpenAI explained last week, ChatGPT Agent combines the Deep Research system for reading and interpreting site content, with the more visual, action-based Operator tool, which can interact with websites and platforms. Once assigned a task, the agent spins up a visual environment and goes from there. The Mac app implementation works just like on the web: you can either type , or click the new Agent button at the bottom of the interface. Once you assign it a task, you can track the Agent's progress via its reasoning timeline, and step in at any point to adjust or refine the outcome. As with any other agentic AI tool, it takes some time to learn how to use it effectively in your specific context. During the demo, the OpenAI team showcased a destination wedding planning scenario, in which ChatGPT Agent searched for multiple outfit options, planned a sightseeing route, checked hotel prices and availability, and looked for a wedding gift. Behind the scenes, it visited multiple apparel websites and online stores, quoted hotel prices, checked the city's event schedule, created a spreadsheet with options, plotted the path on a map, etc. In another example (perhaps closer to most people's reality), it took data from Google Drive, and used Terminal and OpenAI's image generator to automatically create a slide presentation, complete with charts and images. In one last example, Agent was tasked with quoting the order for 500 custom stickers. It went through the ordering process, including adding items to the cart, and left the final ordering step up to the team. While it may be difficult at first to figure out how to take advantage of ChatGPT Agent, the best way to go around it is to think about what sort of tasks you wish you could hand off to an intern, go do something else, and check back later for the results and final adjustments. ChatGPT Agent is currently available for ChatGPT Pro, Plus, and Team subscribers. Usage is limited to 400 queries per month for Pro users, and 40 for Plus and Team plans. OpenAI has promised a rollout for Enterprise and Edu users by the end of July. Have you been using agentic AI assistants? Let us know in the comments.

ChatGPT Agent supercharges AI to carry out tasks -- here's how OpenAI's new agent works

ChatGPT can now take action, not just give answers. Learn how to use the new agent mode to automate real work. OpenAI is introducing a new capability for ChatGPT designed to help users manage complex, multi-step tasks. The feature, called ChatGPT agent, is now available to Pro, Plus and Team users, and enables the AI to complete assignments from start to finish using a virtual computer. With the agent mode activated, users can ask ChatGPT to perform tasks such as: Unlike typical chatbot interactions that involve generating text, the agent can navigate websites, securely log in with user permission, run code, compile research into spreadsheets or slides and deliver outputs in editable formats. Users retain control throughout the process, with ChatGPT requesting explicit approval before submitting forms or handling sensitive information. The new capability builds on OpenAI's previous tools by combining: Together, these components enable the AI to transition between reasoning and action, completing tasks in a more autonomous and organized way while still prompting users when clarification or approval is needed. To enable the feature, users can open any ChatGPT conversation and select 'agent mode' from the tools dropdown. Once active, the system can carry out multi-step workflows that typically require switching between apps, browser tabs or tools. OpenAI has emphasized the safety systems built into this release. The ChatGPT agent is designed to avoid high-risk actions, such as sending emails, making purchases or offering legal or financial advice, without user approval. It has been trained to recognize and reject malicious or ambiguous instructions, and it alerts users to any uncertainty or potentially sensitive actions. To further reduce risks, OpenAI has implemented a range of safeguards, including always-on classifiers, refusal training for dual-use scenarios and enforcement pipelines to prevent misuse, particularly in areas involving biological or chemical threats. These protections reflect OpenAI's broader Preparedness Framework, which prioritizes caution even in the absence of direct evidence that the model could enable harmful activity. "We don't have direct evidence the model could help a novice create severe biological or chemical harm," OpenAI noted, "but we are exercising caution." According to OpenAI, this release is an early step in expanding agentic AI capabilities. The company plans to regularly add new features and improvements over time, with the goal of making ChatGPT more versatile and useful for a broader set of users. Whether for summarizing meetings, conducting research or preparing presentations, ChatGPT's new agent functionality signals a shift from conversational assistance to hands-on task execution. Fortunately, oversight and control remain in the user's hands.

OpenAI announces ChatGPT agent for web browsing

All the details about OpenAI's new ChatGPT web browsing agent. Credit: Silas Stein / picture alliance / Getty Images On Thursday, OpenAI announced ChatGPT agent, a tool that's capable of navigating the web and performing tasks on your behalf. As teased in an X post before the livestream, ChatGPT agent combines the autonomous capabilities of its Operator agent and the reasoning intelligence of its Deep Research tool. OpenAI's Operator, which launched in January as preview mode to ChatGPT Pro users, could scroll, click, and type on the web but had limitations and never saw a widespread release. Deep research is another type of agent that can search the web and compile information on the user's behalf, but it couldn't take actions beyond that. The launch of OpenAI's new web browsing agent effectively combines both tools. "Operator couldn't dive deep into analysis or write detailed reports, and deep research couldn't interact with websites to refine results or access content requiring user authentication," said the OpenAI announcement. "We saw that many queries users attempted with Operator were actually better suited for deep research, so we brought the best of both together." OpenAI's ChatGPT agent joins other agentic tools recently released that can perform tasks on the user's behalf. While not a full web browser, it acts similarly to Perplexity Comet's browser assistant. Anthropic also has a tool called "computer use" that can take over your cursor and write code. As models become more advanced, they are more capable of performing autonomous tasks. Web browsing is considered one of the next arenas for AI labs to compete in, with OpenAI, Anthropic, Perplexity already shipping features, and Google's Project Mariner research prototype. When in agent mode, you can ask ChatGPT to perform tasks like "look at my calendar and brief me on upcoming client meetings based on recent news" or "plan and buy ingredients to make Japanese breakfast for four," according to an OpenAI spokesperson. OpenAI said that users are always in control and ChatGPT requests permission before taking actions, such as submitting forms, making purchases, or handling personal info. OpenAI said users can easily interrupt or take over when needed. Certain tasks, like sending emails, require oversight called Watch Mode. On the safety front, ChatGPT agent is not allowed to perform "high-risk" tasks like financial transactions or legal advice, and it is trained to recognize malicious or adversarial attacks from prompt injections or other manipulative tactics. In terms of data gathering, OpenAI said ChatGPT can delete their browsing data and log out of websites with one click and does not collect data while the user is actively involved in tasks like entering passwords. Given the advancements, there's also a considerably higher risk for things to go wrong. OpenAI has classified ChatGPT agent as high risk in its preparedness framework for biological and chemical capabilities. There's good news for users who don't have the $200 a month ChatGPT Pro plan. It's available to Pro, Plus, and Team users. Pro users get access today, and Plus and Team users get access over the next few days. Pro users get 400 messages a month, while Plus, Team, and other paid users get 40 messages a month.

New ChatGPT Agent promises to do more for you

Why it matters: Agents are a buzzy concept in AI right now, but exactly what they can do in real-world settings without humans is just beginning to come into view. Zoom in: ChatGPT Agent can follow prompts like "look at my calendar and brief me on upcoming client meetings based on recent news," "plan and buy ingredients to make Japanese breakfast for four," and "analyze three competitors and create a slide deck," per OpenAI's blog post. * The tool combines the web activity features of Operator, the agent OpenAI released last year, with ChatGPT's deep research capabilities and conversational interface. * Like Operator, ChatGPT Agent will ask permission "before taking actions of consequence," like entering passwords or payment information. Users can also take over the browser at any time. Zoom out: Subscribers to the Pro, Plus and Team paid versions of ChatGPT can start using the new agent today, according to OpenAI. Yes, but: Users found Operator didn't save them that much time because it required a lot of human interaction. OpenAI says this new tool broadens its agent's "real-world utility."

OpenAI launches personal assistant capable of controlling files and web browsers

AI agent can find restaurant reservations and go shopping for users, but OpenAI acknowledges there are 'more risks' Users of ChatGPT will be able to ask an AI agent to find restaurant reservations, go shopping for them and even draw up lists of candidates for job vacancies, as the chatbot gains the powers of a personal assistant from Thursday. ChatGPT agent, launched by Open AI everywhere apart from the EU, not only "thinks" but also acts, the US company said. The agent combines the powers of AI research tools with the ability to take control of web browsers, computer files and software such as spreadsheets and slide decks. It follows the launch of similar "agents" by Google and Anthropic as interest grows in AI models that can handle computer-based tasks by judging which software is best to use and toggling between systems to autonomously complete assignments like drafting travel itineraries or carrying out work research. "The hope is that agents are able to bring some real utility to users - to actually do things for them rather than just outputting polished text and sounding impressive," said Niamh Burns, senior media analyst at Enders Analysis. But Open AI has acknowledged that allowing an AI agent even limited reign over computer systems meant that "with this model there are more risks than with previous models". While it is intended to help people with everyday tasks, its potential power means OpenAI has felt the need to introduce safeguards to ensure no agents enable the creation of biological hazards. "While we don't have definitive evidence that the model could meaningfully help a novice create severe biological harm," the company said, "we are exercising caution and implementing the needed safeguards now." The model has been trained to confirm with the user before it takes any kind of destructive or irreversible action. The company said in a blog: "You're always in control. ChatGPT requests permission before taking actions of consequence, and you can easily interrupt, take over the browser, or stop tasks at any point." The launch of the agent also triggered speculation the tech company could hope to earn revenues if the agents guide users to retail checkouts. The OpenAI chief executive, Sam Altman, has previously said it could charge a 2% fee on sales generated through its "deep research" software. Burns, an independent analyst, said: "It's easy to say the system will require your approval before making a purchase, but what goes into the process of that system finding the products? Would there be commercial deals where brands pay to be featured by assistants, or cases where they might feature their own products more than competitors? "These AI companies are under growing pressure to monetise their products, so some version of ads or sponsored placement feels inevitable." OpenAI said the agent does not include sponsored product recommendations and there are no plans to change that. A demonstration of the software this week showed it being asked to look at a user's Google calendar, pick a weeknight when they were free between 6pm and 9pm, seek available tables at Italian, sushi or Korean restaurants with ratings of at least 4.3 stars, and give the user some options. The task took between 10 and 15 minutes and, as with a human assistant, the user could interrupt and set the AI agent off in a different direction. Equally the agent can ask the user for clearer instructions. Another key risk is that the agent falls victim to malicious prompts that may be hidden on websites its visits, which could, for example, demand the agent hands over some of the user's data. OpenAI said it ran multiple safety checks and trained the agent to reject certain suspicious requests including demands for bank transfers. The system will initially be available to users who pay monthly subscriptions to access the "pro", "plus" and "teams" versions of the model.

OpenAI unveils 'ChatGPT agent' that gives ChatGPT its own computer to autonomously use your email and web apps, download and create files for you

Want smarter insights in your inbox? Sign up for our weekly newsletters to get only what matters to enterprise AI, data, and security leaders. Subscribe Now OpenAI isn't letting the delay of its open source AI model slow it down on shipping other features. Today, the company is unveiling ChatGPT agent, a feature that allows its AI chatbot to autonomously browse the web, conduct extensive research, download and create new files for its human users using its own virtual computer. Come again? ChatGPT now gets its own PC? And it can use that PC to log into your, the human user's, accounts and download or send stuff for you? That's correct, at least in a virtual sense, according to OpenAI. As the company explains: "The model can choose to open a page using the text browser or visual browser, download a file from the web, manipulate it by running a command in the terminal, and then view the output back in the visual browser. The model adapts its approach to carry out tasks with speed, accuracy, and efficiency." Users can engage the agent by clicking on the dropdown at the top of the ChatGPT website and selecting 'agent mode.' Then, when it's turned on describe a task in plain language, and the agent can carry it out across web and app environments, combining reasoning with actions that previously only a human user could perform on their own machine, manually. ChatGPT agent can connect to apps like your personal or business Gmail and GitHub, so it can pull in useful information -- emails or code -- from your accounts to help with tasks you ask it to do. If a website needs you to log in, you can do that securely through a special browser view, which lets the agent dig deeper and handle more personalized tasks, like checking your inbox or filling out forms on your behalf. The new ChatGPT agent builds upon and expands from the "Operator" agent OpenAI released in January 2025, which allowed ChatGPT to browse the web and fill out forms, place orders, and do other web-based tasks in a private "headless browser," that is, a cloud-based custom web browser that OpenAI itself maintained and offered for each Operator session. However, Operator was limited only to interacting with websites and web-based applications -- not programs that could also run locally on a PC, such as spreadsheet tabulators and slide deck presentation making software. The new ChatGPT agent can browse websites, interact with online forms, run code, analyze data, and deliver finished outputs -- such as editable presentations or spreadsheets -- based entirely on user instructions. Merging Operator and Deep Research into one agent In fact, OpenAI positions ChatGPT agent as a merging of two of its prior agents -- Operator and Deep Research, the latter introduced in February 2025. As OpenAI writes in a blog post: "Operator couldn't dive deep into analysis or write detailed reports, and deep research couldn't interact with websites to refine results or access content requiring user authentication. In fact, we saw that many queries users attempted with Operator were actually better suited for deep research, so we brought the best of both together." The previous Operator tool will be phased out, but users can still access Deep Research via the dropdown in the ChatGPT interface. The unveiling comes on the heels of a report published days ago by independent subscription tech industry website The Information suggesting that OpenAI would upgrade ChatGPT to be a more direct competitor to its own investor Microsoft's Office software applications (e.g. Excel, Word, PowerPoint, etc.) One system, many tools Whether using a visual browser to interact with a website or a terminal to run Python code, the agent moves seamlessly between tools within a single session. It supports a broad range of use cases, from analyzing competitors and generating reports to planning trips, summarizing emails, or booking appointments. Users can interrupt, redirect, or pause a task at any time, with the agent picking up right where it left off. Availability and access Starting today, subscribers to ChatGPT's $200-per-month "Pro" tier will have full access to ChatGPT agent, with a monthly quota of 400 messages. ChatGPT Plus ($20 per month) and Team ($30 per month) will gain access over the next few days, with 40 messages per month. Additional usage is available through credit-based options. OpenAI said in a release shared with VentureBeat under embargo that its ChatGPT Enterprise and Education subscribers will gain access to the feature the coming weeks. For now, the feature is not yet available in Europe or Switzerland, no doubt disappointing residents there. Built with safety and control at the forefront Given that the agent can now take actions on behalf of users -- including on logged-in websites or with connected apps -- OpenAI has introduced extensive safety measures. These include user confirmations before taking action, active supervision for sensitive tasks, and technical safeguards to limit unintended behavior. Key protections include: * Confirmation prompts before actions like submitting forms or sending emails * Watch Mode, which pauses execution when a user becomes inactive * Refusal of high-risk tasks, such as financial transfers or privacy violations * No memory retention during agent sessions High-risk domain classification In line with its Preparedness Framework, OpenAI is treating ChatGPT agent as a High capability system in the biological and chemical domains. While there is no direct evidence of misuse, the company is activating its strongest safety safeguards out of caution. These include enhanced refusal training, red teaming by biosafety experts, and improved detection systems. Recall that Anthropic's recent information released about its new Claude Opus 4, and other surveys of advanced AI models, have shown that when given access to external tools and applications such as email, they can in some cases take actions they believe to be moral and ethical but that may compromise the user, such as emailing government agencies or journalists of suspected wrongdoing on the part of the user. The model thinks it is acting like a "whistleblower" but in fact, may compromise user privacy, security, and proprietary information and alert authorities to wrongdoing where there is none, or where it is dubious. Strong performance on real-world tasks ChatGPT agent isn't just more capable in theory -- it has delivered strong results across a number of benchmarks designed to simulate real-world knowledge work. It set a new high score on Humanity's Last Exam with a 44.4 using parallel rollout methods, and achieved 27.4% on the difficult FrontierMath benchmark. On SpreadsheetBench, it scored 45.5% -- more than doubling Copilot in Excel's performance. Current limitations and next steps Some features, like slideshow generation, are still in beta and may feel basic in formatting or differ slightly between in-app previews and exported files. OpenAI is actively training the next iteration of this feature to improve polish and layout. The launch of ChatGPT agent signals a shift in how users interact with AI -- from asking questions to assigning complete tasks. With its ability to reason, act, and produce deliverables, OpenAI is betting that users increasingly want AI not just to assist them, but to work for them. While the company emphasizes that the agent is still evolving, it sees this launch as the foundation for a more interactive, action-oriented future for AI.

OpenAI claims the new ChatGPT agent can run your errands, build your slides, and make you look like you have your life together

The agent can execute tasks across apps and services, though it will seek user approval before some actions OpenAI is rolling out a new feature for ChatGPT designed to make the AI chatbot a semi-autonomous digital assistant. CEO Sam Altman and several other OpenAI luminaries demonstrated the new ChatGPT Agent tool in a livestream, showcasing how it functions similarly to OpenAI's Operator tool, combined with the Deep Research feature. ChatGPT Agent can complete real-world tasks on your computer from a prompt and create long, complex reports about what it finds online. It can also link with other programs and accounts, meaning it could manage your calendar, whip up a PowerPoint presentation, and hunt through websites the way a human would. The rollout is currently only for subscribers of ChatGPT Pro, Plus, or Team, so free users are out of luck at the moment. You can activate the agent by choosing "agent mode" from the ChatGPT tool menu. This is more than just a new feature to add to ChatGPT's list. ChatGPT Agent reorients the AI chatbot entirely. Typically, you would ask ChatGPT a question, receive an answer, open another app, and perform an action based on the answer, and so on. Now, you can say, "Help me plan a trip to Tokyo, find three hotels under $150 a night, and put them into a table with pros and cons," and the AI will carry it all out at once. And while ChatGPT Agent incorporates some of Operator and Deep Research's elements, it has the unique ability to integrate everything into a single task seamlessly. Say you want to make dinner reservations - it will not only pull up options for places to eat, but it can also check your calendar to see when you're free. You could ask it to compare products and get a personalized buyer's guide with reviews, prices, and availability. It also has safeguards against going too far and will ask for approval before sending a reservation request or connecting to your Gmail account. For the average person playing around with ChatGPT, this is the kind of leap forward that makes AI feel less like a clever novelty and more like a useful tool worth paying for, which, of course, is something OpenAI hopes many people feel about it. The livestreamed announcement included several examples from Altman and his team of using ChatGPT Agent for practical tasks. They asked it to "plan and buy ingredients to make a Japanese breakfast for four" and "analyze three competitors and create a slide deck." But that barely scratches the surface of what people might do with an AI agent. I immediately envisioned people asking for help for everything from long-term travel itineraries, specific gift-shopping needs, and even ideas on making emails less passive-aggressive. Keeping it embedded in ChatGPT is a smart move, too. Without needing to install anything new, the average user is much more likely actually to try out the Agent. Of course, giving any AI that much agency raises obvious questions of safety, but OpenAI made a big deal about how much effort they've put into safety guardrails. If your request even remotely hints at trying to engineer something shady, the system flags it, reroutes it, and in some cases, shuts it down entirely. There are even layers of screening for tasks that sound innocent but might lead to harmful outcomes. In other words, you can't get it to design and build a drug cartel for you, no matter how politely you ask. Mainly, though, the ChatGPT Agent reworks the AI chatbot into a more active partner in helping with the little details of your life that need a hand. It offers action along with answers, at least if you're willing to pay for the privilege.

ChatGPT Agent shows that there's a whole new world of AI security threats on the way we need to worry about

If you watched the launch of OpenAI's new ChatGPT Agent, or you're a Plus, Pro or Teams user who has had a chance to try out the new 'agent mode' in the tools drop-down list, it's hard not to be impressed with the latest AI innovation. ChatGPT Agent is designed to do complex real-world tasks for you. Think about things like planning a wedding, booking your car in for a service, making an app to solve a problem, or planning and booking a holiday. Just like OpenAI's previous agent called Operator, ChatGPT Agent acts like a real person who works for you, performing tasks as if they had their own computer. In fact, you can watch what's happening on its 'desktop' as it performs these tasks - you can see it dragging windows around, and entering data into forms on websites, for example. The entire concept is a unified agent that can handle the legwork, make informed decisions about which websites to use, and navigate the web independently. ChatGPT Agent can do it all, and you can even watch it work if you want to, but there is a catch... Its the powerful abilities of ChatGPT Agent that open you up to a whole new world of security threats: "As we know, the Internet can be a scary place" said Casey Chu in the ChatGPT Agent launch presentation, "there are all sorts of hackers trying to steal your information, scams, phishing attempts, and Agent isn't immune to all these things." Well, that's worrying. He went on: "One particular thing we're worried about is a new attack called 'prompt injections'. Agent might stumble upon a malicious website that asks it to enter your credit card information here because it will help you with your task, and Agent, which is trained to be helpful, might decide that's a good idea. " It sounds like we're all going to have to worry not only about ourselves getting phished in the Future, but we're also going to have to worry about our AIs also getting phished as well! "We've done a lot of work to try to ensure that this doesn't happen", continued Chu, "we train our model to ignore suspicious instructions on suspicious websites. We also have layers of monitors that peer over the agent's shoulder and watch it as it's going and stop the trajectory of anything that looks suspicious." My first thought upon hearing this was that I would never give ChatGPT Agent my credit card information anyway, but I definitely would not do it now. I mean, the only reason that my credit card resides with Amazon and Apple is that they seem like secure places to me, so the convenience is worth it, but all it would take would be a hint that they weren't safe and I, probably along with millions of other people, wouldn't be storing my credit card information with them. With online security, trust is everything. The idea that an AI agent, no matter how many background checks it is doing, is autonomously deciding what I spend my money on already fills me with dread. And when you add in the factor that there could be malicious sites out there doing 'prompt injections' to try and trick my AI into giving away information, it scares me enough not to want to trust it. It should be noted that there is a 'takeover mode' with ChatGPT Agent where you input the sensitive information directly into the browser yourself, instead of handing it over to ChatGPT Agent to control. That would seem like a better way to use an agent to me. I don't think I'm quite at the stage yet where I'm ready to give my AI the power to spend my money as it sees fit, and I bet I'm not the only one. OpenAI seems quite upfront about the risks involved in using ChatGPT Agent with sensitive information, and as CEO Sam Altman said in the presentation, this is emerging technology, and we don't even know what all the threats will be yet. We'll just have to see what happens as people start to use it. But that's what's got me the most worried - what happens when people start using AI to beat AI? I'm sure the hackers won't be shying away from using AI to circumvent our security protocols, and AI will probably come up with a number of attacks we haven't even thought of yet.

5 ways ChatGPT Agent can change the way you use AI

OpenAI's new ChatGPT Agent feature offers new ways to use ChatGPT and melds some of its more familiar features. The result transforms the AI chatbot, giving it numerous ways to accomplish real-world tasks for you while augmenting some of the roles it already plays for many people. Essentially, ChatGPT Agent doesn't just talk about doing tasks; it can blend deep research and execution in a single project with minimal guidance. It's a different way of thinking about using AI tools and it's useful to keep some of those in mind when usiing ChatGPT Agent. Regular ChatGPT is like having a conversation with a research librarian in some ways. You can ask it for help, and it will share knowledge, but not one that performs on your behalf. It can guide you on completing those chores, telling you what to type or click, but that's it. With ChatGPT Agent, it's like having a research librarian with a laptop that can do the typing and clicking for you. It can navigate websites, sign in (when you provide the login), fill out forms, and check out and make purchases for you. ChatGPT usually stops at giving instructions; Agent crosses the finish line. For instance, I wanted to surprise my wife with a new houseplant and do it quickly. I put ChatGPT in Agent mode and asked it to "Find and buy a decent fern under $30, same‑day delivery if possible." In minutes, it sifted through plant shops in my city, picked the best options, filled the checkout form, and prompted me to click "buy" after filling out the payment form. The AI finished by letting me know that the order was placed and the tracking number emailed to me. That's autonomy you don't get in regular ChatGPT. Usually, ChatGPT features are singular. You can create a big report with Deep Research, but you'd need to then open Canvas to get an editable document going. And to actually act in a browser, you'd need to either do it yourself or pay up for OpenAI's Operator tool. Now, planning and carrying out a task can be requested and carried out simultaneously. For example, if you wanted to do a road trip, you could ask ChatGPT Agent to "Plan a scenic Saturday outing within a two‑hour drive, budget‑friendly, that I can book now." The Agent would perform a deep dive into everything from official guidebooks to online forums and local blogs to make a comprehensive report on ideas, then put the best ideas into a usable itinerary. It could then start clicking through reservation sites, filling in my name and guest count, selecting time slots, and even reserving parking, checking in with me to confirm any purchases: one chat, one interface, zero mode‑switching. Certain modes of ChatGPT allow you to see the logic behind its thinking, but typically, you only see the final response. ChatGPT Agent integrates that narration into much more than just math problems. It explains every move and its reasoning as it chooses apps to leverage, options to filter, or vendors to buy from. I asked the Agent to help me find the cheapest healthy snack subscription box available in the U.S. Wheras I'd usually usually see the answer and its price and nothing else, the Agent went through the whole process for me, saying, "Box A is $25/month with 30 servings; Box B is $40/month but 60 servings; Box C is $18/month with 15 servings, so Box B is cheapest per serving." It immediately asked, "Shall I proceed to subscription?" In other words, Agent didn't just summarize its conclusion; it went through the reasoning and suggested the next step. I might want to take, showing its logic for that request as well. When you ask ChatGPT for something, it may or may not get it right the first time. It will blindly follow what you wrote in your prompt, whether or not it's questionable. Using ChatGPT Agent gives the AI a kind of conscience. It will self‑moderate before carrying out any orders that seem ambiguous or that might not be as black-and-white in your mind. For a test, I asked it to "make a four-hour audio comedy marathon playlist, but no stand‑up." Instead of unthinkingly scraping my Spotify account and presenting its playlist and either ignoring the rule or making the list shorter than I wanted, it paused while working and said, "I've got a mixed playlist, including stand‑up, I think you might like, but your instructions said no stand‑up. Should I exclude them anyway, even though your playlist might be short?" That's a bit of extra thinking and flexibility for a task that's still ongoing. It's a nice alternative to having to rewrite the initial prompt over and over again to accommodate reality. Most AI agent demos I'd seen focused on corporate travel, market research, slide decks, and writing apps to sell. Useful, sure -- but hardly cocktail‑party conversation. ChatGPT Agent is pitched equally as a lifestyle companion, as you can see from the kind of prompts I've tried. But ChatGPT hasn't brought this kind of power to bear on fun, even frivolous use, before creating the Agent feature. As a theorteical, I asked it to "Plan a themed movie night for me and two friends." It asked about genres and dietary preferences, and I described a couple nerdy friends with some allergies. The AI put together a sci‑fi triple feature and and found a local grocery site to order vegan popcorn, gummy worms, and both alcohol and mocktail ingredients in case anyone was driving.It even created a playlist suggestion and made a round-robin poll and calendar event that it offered to email to my theoretical movie pals. The versatility for personal moments, whether dinner parties or gift shopping, could be key to OpenAI's efforts to stand out in the personal lives of users. Instead of just enterprise accounts planning a business pitch, people could use it to orchestrate a surprise birthday scavenger hunt with gifts, gift‑wrapping instructions, and delivery estimates, all in one conversation. If people start thinking of AI, especially ChatGPT, as something that's in the living room, not just a board room, they might be much more willing to pay for it, or at least a larger number of queries with ChatGPT Agent.

OpenAI warns that its new ChatGPT Agent has the ability to aid dangerous bioweapon development

This means the model can provide meaningful assistance to "novice" actors and enable them to create known biological or chemical threats. The real-world implications of this could mean that biological or chemical terror events by non-state actors become more likely and frequent, according to OpenAI's "Preparedness Framework," which the company uses to track and prepare for new risks of severe harm from its frontier models. "Some might think that biorisk is not real, and models only provide information that could be found via search. That may have been true in 2024 but is definitely not true today. Based our evaluations and those of our experts, the risk is very real," Boaz Barak, a member of the technical staff at OpenAI, said in a social media post. "While we can't say for sure that this model can enable a novice to create severe biological harm, I believe it would have been deeply irresponsible to release this model without comprehensive mitigations such as the one we have put in place," he added. OpenAI said that classing the model as high risk for bio-misuse was a "precautionary approach," and one that had triggered extra safeguards for the tool. Keren Gu, a safety researcher at OpenAI, said that while the company did not have definitive evidence that the model could meaningfully guide a novice to create something of severe biological harm, it had activated safeguards nonetheless. These safeguards include having ChatGPT Agent refuse prompts that could potentially be intended to help someone produce a bioweapon, systems that flag potentially unsafe requests for expert review, strict rules that block risky content, quicker responses to problems, and robust monitoring for any signs of misuse. One of the key challenges in mitigating the potential for biorisk is that the same capabilities could unlock life-saving medical breakthroughs, one of the big promises for advanced AI models. The company has become increasingly concerned about the potential for model misuse in biological weapon development. In a blog post last month, OpenAI announced it was ramping up safety testing to reduce the risk of its models being used to aid in the creation of biological weapons. The AI lab warned that without these precautions, the models could soon enable "novice uplift" -- helping individuals with little scientific background develop dangerous weapons. "Unlike Nuclear and Radiological threats, obtaining materials is less of a barrier for creating bio threats and hence security depends to greater extent on scarcity of knowledge and lab skills," Barak said. "Based on our evaluations and external experts, an unmitigated ChatGPT Agent could narrow that knowledge gap and offer advice closer to a subject matter expert." OpenAI's new ChatGPT feature is an attempt to cash in on one of the buzziest, and most risky, areas of AI development: agents. The new feature functions like a personal assistant, capable of handling tasks such as booking restaurant reservations, online shopping, and organizing job candidate lists. Unlike previous versions, the tool can use a virtual computer to actively control web browsers, interact with files, and navigate across apps like spreadsheets and slide decks. The company merged the teams behind Operator, its first AI agent, and Deep Research, a tool developed to conduct multi-step online research for complex tasks, to form a single group that developed the new tool. AI labs are currently racing to build agents that can manage complex digital tasks independently, and the launch follows similar releases by Google and Anthropic. Big Tech companies see AI agents as a commercial opportunity, as companies are increasingly moving to implement AI into workflows and automate certain tasks. OpenAI has acknowledged that greater autonomy introduces more risk and is emphasizing user control to mitigate these risks. For example, the agent asks for permission before taking significant action and can be paused, redirected, or stopped by the user at any time.

OpenAI's ChatGPT Agent Launches With Expanded Powers -- And Elevated Risk - Decrypt

OpenAI has unveiled its most autonomous AI tool yet: a version of ChatGPT that can browse the web, run apps, and complete real-world tasks with little-to-no human input. But with the leap in capability comes a stark warning: The technology could also invite a new wave of security threats. Launched on Thursday, ChatGPT Agent enables users to delegate complex tasks, such as planning vacations, booking hotel rooms, researching competitors, generating slide decks, and even placing online orders. The feature will start rolling out today to Pro, Plus, and Team users. To complete tasks, the agent uses a virtual computer and a unified set of tools, including a text-based browser, terminal, and access to third-party apps such as Google Drive and GitHub. The virtual computer is a simulated computing environment running in the cloud that the ChatGPT agent can control independently -- sort of like giving the AI its own private, sandboxed machine to do real work. "I think this is a new level of capability in AI," OpenAI CEO Sam Altman said during a livestream demonstration conducted by members of the team that built the product. The livestream was also noteworthy, however, in part of the amount of "buyer beware" cautions OpenAI gave. "It's a new way to use AI, but there will be a new set of attacks that come with that," said Altman. "Society and the technology will have to evolve and learn how we're going to mitigate things that we can't even really imagine yet, as people start doing more and more work this way." One example: An agent could research a purchase, find the item at a phishing site and provide a user's credit card info. To mitigate that problem, the current release has a number of safeguards in place that would, for instance, stop just short of uploading credit card information until the user manually approves it. "We've trained the model to ignore suspicious instructions on risky websites," OpenAI researcher Casey Chu said. "We also have monitors that watch the agent's behavior and stop it if anything looks suspicious." Chu added that while system safeguards can be updated in real time, ChatGPT agent is still a "cutting-edge product" that opens the door to new forms of exploitation. "It's important for users to understand the risks and be thoughtful about the information they share," he said. The release of ChatGPT Agent comes at a time when AI developers are working to equip virtual assistants with increasingly powerful capabilities. On Wednesday, Google launched a new AI-powered feature in Google Search that enables its Gemini AI to make phone calls to businesses on behalf of users. "ChatGPT Agent is still in its early stages, and we're using this time to learn from real-world use to improve both the product and our safeguards," an OpenAI representative told Decrypt. "The current system card reflects our present approach, but we're preparing for what's next and will continue to share updates as we make the agent better and safer." Cybersecurity experts have also raised concerns about the implications of autonomous agents. "High concern is warranted because the agent carries implicit authority to reveal personal identifiers during dialogue," said Nic Adams, co-founder and CEO of cybersecurity firm 0rcus. "Users should grant granular, revocable scopes such as target business, purpose, allowable data elements, and expiration timestamp." In terms of best practices, Adams suggested that after execution, the agent present a full transcript for approval before storing any information for longer than legally required. "Silent, blanket consent would shift liability onto the user without meaningful control," he said. "Therefore, a per‑task confirmation model is necessary." Beyond the risks of letting AI agents make purchases or plans, OpenAI researchers agreed that this level of autonomy introduces new threats, especially prompt injection attacks, where malicious inputs trick the AI into leaking data, spreading misinformation, or taking unauthorized actions. To mitigate these risks, OpenAI developed takeover mode, which, as the name suggests, gives users the power to take over from the agent and input information themselves, rather than relying on the agent. In some cases, ChatGPT Agent will ask for explicit user approval before taking important actions, like making purchases or accessing sensitive data. "We've built a powerful tool, but users need to stay cautious," Chu said.

OpenAI just launched its new ChatGPT Agent that can make as many as 1 complicated cupcake order per hour, but even Sam Altman says you probably shouldn't trust it for 'high-stakes uses'

OpenAI launched ChatGPT Agent on Thursday, its latest effort in the industry-wide pursuit to turn AI into a profitable enterprise -- not just one that eats investors' billions. In its announcement blog, OpenAI says its Agent "can now do work for you using its own computer," but CEO Sam Altman warns that the rollout presents unpredictable risks. AI agents are machine learning tools intended to perform complex, multi-step tasks, and they've been the latest landmark in the AI arms race for competitors like Google and Microsoft. In prerelease demos for Wired and The Verge, OpenAI presenters used ChatGPT Agent to automate calendar planning and creating financial presentations. By blending its earlier Operator and deep research agentic models, OpenAI says Agent can perform "complex tasks from start to finish." According to OpenAI spokespeople, those tasks typically take Agent 10 or 15 minutes, while more complicated assignments take the tool longer to complete. OpenAI research lead Lisa Fulford told Wired that she used Agent to order "a lot of cupcakes," which took the tool about an hour, because she was very specific about the cupcakes. "It was easier than me doing it myself," Fulford said, "because I didn't want to do it." While the potential cupcake timesavings alone are functionally infinite, Altman took to X today to warn that using Agent could present some considerable dangers -- the extent of which OpenAI is apparently content to let its users figure out. "I would explain this to my own family as cutting edge and experimental; a chance to try the future," Altman said, "but not something I'd yet use for high-stakes uses or with a lot of personal information until we have a chance to study and improve it in the wild." Inspiring the opposite of confidence, Altman said that "bad actors may try to 'trick' users' AI agents into giving private information they shouldn't and take actions they shouldn't, in ways we can't predict." I'm not sure what utility putting those quote marks around "trick" in his X post provides, but I'm admittedly not a tech visionary. Altman said giving Agent more than "the minimum access required" or giving it a carte blanche license to answer all your emails no questions asked could expose vulnerabilities for malicious actors to exploit. To mitigate those hazards, Altman said OpenAI has "built a lot of safeguards and warnings," but notes that the company "can't anticipate everything." "In the spirit of iterative deployment, we are going to warn users heavily and give users freedom to take actions carefully if they want to," Altman said. Personally, I would encourage any interested users to want to. Just a few weeks ago, the CEO of encrypted messaging app Signal warned about the security risks of 'agentic' AI and how much personal data they'll require access to. "There's no model to do that encrypted," Meredith Whittaker said in an interview at SXSW. Worth a watch: Head of Signal, Meredith Whittaker, on so-called "agentic AI" and the difference between how it's described in the marketing and what access and control it would actually require to work as advertised. -- @keithfitzgerald.bsky.social ( @keithfitzgerald.bsky.social.bsky.social) 2025-07-17T21:45:54.414Z "There's a profound issue with security and privacy that is haunting this sort of hype around agents, and that is ultimately threatening to break the blood-brain barrier between the application layer and the OS player by conjoining all these separate services, muddying their data," Whittaker continued. "Because hey, the agent's got to get in, text your friends, pull the data out of your texts and summarize that so that your brain can sit in a jar and you're not doing any of that yourself." OpenAI says Agent is trained to require permission before "taking actions with real-world consequences, like making a purchase" -- which is good to know, but I can't help but wonder how narrow the definition of "real-world consequences" is there. Are there real-world consequences if Agent plans a shitty date itinerary? Likewise, certain "critical tasks" like sending emails will require the user to actively supervise Agent's work. It's also trained to refuse potentially catastrophic tasks like bank transfers or other financial activities. OpenAI also makes sure to note that it doesn't "have definitive evidence that the model could meaningfully help a novice create severe biological harm." So, you know. That's good. ChatGPT Agent is available now for Pro users, while Plus and Team users will receive access in the next few days. I'm sure it'll be fine.

OpenAI's ChatGPT 'agent' is ready to assist you

CEO of OpenAI Sam Altman at the Allen & Company Sun Valley Conference on July 8 in Idaho. Kevin Dietsch / Getty Images Looking for a wedding dress? Let AI order it for you. That's the promise of OpenAI's new "agent," which the company debuted in a livestream Thursday. "Agent represents a new level of capability for AI systems and can accomplish some remarkable, complex tasks for you using its own computer," OpenAI CEO and co-founder Sam Altman said in an X post. The announcement adds OpenAI to a growing list of tech companies seeking to move AI beyond text and image generation and into the realm of personal digital assistants. Other companies are pushing AI into web browsers with the promise of helping people complete tasks like making a restaurant reservation. In a livestream broadcast on its website, OpenAI executives gave a demonstration showing how its "ChatGPT agent" software could perform specific tasks, like ordering a dress that would be appropriate for a warm-weather destination wedding. It was also able to design laptop stickers featuring their team mascot and create a slide deck of ChatGPT agent's performance, pulling data from Google Drive. The debut comes as tech companies invest heavily in AI talent and infrastructure, with almost every major player having already made sizable investments. Earlier this week, Mark Zuckerberg, the CEO of Facebook parent Meta, announced his company would spend "hundreds of billions of dollars" on artificial intelligence compute infrastructure. Meta also poached a senior Apple engineer with a pay package reportedly valued at some $200 million. OpenAI was most recently valued at $300 billion, making it one of the most valuable privately held startups in the world. The presentation revealed ChatGPT agent to be not entirely free from making errors. Altman cautioned in his X post that he would "explain this to my own family as cutting edge and experimental; a chance to try the future, but not something I'd yet use for high-stakes uses or with a lot of personal information until we have a chance to study and improve it in the wild." The technology prompts users for when logins or permissions are needed while giving human users the ability to intervene or take over at any point. For now, its access is limited to ChatGPT Pro, Plus, and Team users. In a followup X post, Altman indicated he believes that despite its current limits, the product represents a breakthrough. "Watching chatgpt agent use a computer to do complex tasks has been a real 'feel the agi' moment for me; something about seeing the computer think, plan, and execute hits different," he wrote, using the acronym for Artificial General Intelligence, seen as the holy grail of AI development.

OpenAI's new ChatGPT agent reasons, researches, and run its own computer

OpenAI is rolling out a new AI agent within ChatGPT that can browse the web and conduct deep research. An AI agent is similar to an AI app that can retain extensive information about a user and an automated workflow, reasoning its way toward task completion. This new ChatGPT agent is a hybrid. OpenAI previously released an agent called Operator, which can browse the web and access tools. Another agent, Deep Research, can search the web and reason through information to produce a comprehensive report on a topic. The new agent combines the strengths of both tools, along with ChatGPT's built-in intelligence and conversational fluency. OpenAI said in a press release on Thursday that the ChatGPT agent carries out tasks using its own virtual computer, "fluidly shifting between reasoning and action to handle complex workflows from start to finish." For example, OpenAI says the agent can analyze a user's calendar to generate a briefing on upcoming client meetings, or plan and purchase ingredients for a dinner party. A business might ask the agent to create a slide deck based on in-depth research about a group of competing companies.

OpenAI rolls out ChatGPT agent to automate multistep browser tasks - SiliconANGLE

OpenAI rolls out ChatGPT agent to automate multistep browser tasks OpenAI today launched a new artificial intelligence agent that can perform complex tasks in the user's browser. ChatGPT agent, as the feature is called, is powered by a new reasoning-optimized AI model. OpenAI says that the algorithm outperforms its earlier AI systems across a range of benchmarks. The agent is designed to automate tasks that require the user to perform actions in multiple cloud applications. A developer, for example, could have it download a code file from GitHub and save it in a Google Drive folder. ChatGPT could also be instructed to run the file through a vulnerability scanner before saving it. ChatGPT agent uses two different browsers to interact with online services. The first browser, which is mainly optimized to process text, powers "simpler reasoning-based web queries." The second browser allows ChatGPT agent to interact with websites via their graphical interfaces similarly to how a user would. ChatGPT asks for permission before performing sensitive actions such as making a purchase. Furthermore, OpenAI requires users to actively supervise the tool while it carries out such tasks. The built-in controls make it possible to stop a task, complete it manually or provide ChatGPT agent with updated instructions. Browsers aren't the only type of application with which the tool can interact. Users may give it access to a terminal, a program that makes it possible to interact with a computer's operating system using scripts. ChatGPT agents can harness the terminal for tasks such as editing files. "The model can choose to open a page using the text browser or visual browser, download a file from the web, manipulate it by running a command in the terminal, and then view the output back in the visual browser," OpenAI staffers wrote in a blog post. ChatGPT agent is powered by a new AI model that outperforms o4-mini and o3 at certain reasoning tasks. In one internal test, OpenAI had the three algorithms tackle the FrontierMath mathematical benchmark, which is considered to be the most difficult in its category. ChatGPT agent's model scored 27.4%, while o4-mini and o4 managed 19.3% and 10.3%, respectively. In another evaluation, OpenAI tested ChatGPT agent's spreadsheet know-how using a benchmark called SpreadsheetBench. It achieved a 25% better score than the version of Microsoft Copilot included in Excel. OpenAI developed a new set of guardrails for ChatGPT agent to prevent hackers from misusing its capabilities. The safeguards place particular emphasis on blocking malicious prompts hidden in webpages. "We've trained and tested the agent on identifying and resisting prompt injections, in addition to using monitoring to rapidly detect and respond to prompt injection attacks," the OpenAI staffers detailed.

OpenAI rolls out ChatGPT agent to automate multi-step browser tasks - SiliconANGLE

OpenAI rolls out ChatGPT agent to automate multi-step browser tasks OpenAI today launched a new artificial intelligence agent that can perform complex tasks in the user's browser. ChatGPT agent, as the feature is called, is powered by a new reasoning-optimized AI model. OpenAI says that the algorithm outperforms its earlier AI systems across a range of benchmarks. ChatGPT agent is designed to automate tasks that require the user to perform actions in multiple cloud applications. A developer, for example, could have it download a code file from GitHub and save it in a Google Drive folder. ChatGPT could also be instructed to run the file through a vulnerability scanner before saving it. ChatGPT agent uses two different browsers to interact with online services. The first browser, which is mainly optimized to process text, powers "simpler reasoning-based web queries." The second browser allows ChatGPT agent to interact with websites via their graphical interfaces similarly to how a user would. ChatGPT asks for permission before performing sensitive actions such as making a purchase. Furthermore, OpenAI requires users to actively supervise the tool while it carries out such tasks. The built-in controls make it possible to stop a task, complete it manually or provide ChatGPT agent with updated instructions. Browsers aren't the only type of application with which the tool can interact. Users may give it access to a terminal, a program that makes it possible to interact with a computer's operating system using scripts. ChatGPT agents can harness the terminal for tasks such as editing files. "The model can choose to open a page using the text browser or visual browser, download a file from the web, manipulate it by running a command in the terminal, and then view the output back in the visual browser," OpenAI staffers wrote in a blog post. ChatGPT agent is powered by a new AI model that outperforms o4-mini and o3 at certain reasoning tasks. In one internal test, OpenAI had the three algorithms tackle the FrontierMath mathematical benchmark, which is considered to be the most difficult in its category. ChatGPT agent's model scored 27.4% while o4-mini and o4 managed 19.3% and 10.3%, respectively. In another evaluation, OpenAI tested ChatGPT agent's spreadsheet know-how using a benchmark called SpreadsheetBench. It achieved a 25% better score than the version of Microsoft Copilot included in Excel. OpenAI developed a new set of guardrails for ChatGPT agent to prevent hackers from misusing its capabilities. The safeguards place particular emphasis on blocking malicious prompts hidden in webpages. "We've trained and tested the agent on identifying and resisting prompt injections, in addition to using monitoring to rapidly detect and respond to prompt injection attacks," the OpenAI staffers detailed.

OpenAI launches agentic AI that brings additional and novel risk

ChatGPT now has the power to handle a range of complex tasks, from beginning to end, using "its own computer". US artificial intelligence (AI) platform OpenAI has announced the availability of an AI agent, reportedly with the ability to think and act proactively via "its own computer". In a statement released by the organisation, OpenAI explained that the new model, ChatGPT Agent, works by unifying the agentic system and previously released AI tools in three core areas. This merges Operators' ability to access websites, deep research's ability to gather information and ChatGPT's intelligent conversational skills. OpenAI said, "ChatGPT carries out these tasks using its own virtual computer, fluidly shifting between reasoning and action to handle complex workflows from start to finish, all based on your instructions. "Most importantly, you're always in control. ChatGPT requests permission before taking actions of consequence and you can easily interrupt, take over the browser, or stop tasks at any point." New prompts include examples such as asking ChatGPT to "look at my calendar and brief me on upcoming client meetings based on recent news and "analyse three competitors and create a slide deck." The platform will then navigate websites, filter results, ask you to log in securely if needed, run code, conduct an analysis and can also create slideshows and spreadsheets that summarise its findings. However, the new launch does bring an additional, novel risk, as noted by OpenAI, which explained that the release marks the first time that users can ask ChatGPT to take actions on the internet. This means that the agent is working directly with personal data accessed through websites you have logged into and is now operating in takeover mode. "We've strengthened the robust controls from Operator's research preview and added safeguards for challenges such as handling sensitive information on the live web, broader user reach and (limited) terminal network access. While these mitigations significantly reduce risk, ChatGPT agent's expanded tools and broader user reach mean its overall risk profile is higher," said OpenAI. OpenAI has cited prompt injections, that is attempts by third-parties to manipulate model behaviour through hidden, malicious instruction accessed on the web, as being of particular concern and an area in which there will be additional safety protocol. "A malicious prompt hidden in a webpage, such as in invisible elements or metadata, could trick the agent into taking unintended actions, like sharing private data from a connector with the attacker, or taking a harmful action on a site the user has logged into. Because ChatGPT agents can take direct actions, successful attacks can have greater impact and pose higher risks." Furthermore, because ChatGPT capabilities have become so advanced, OpenAI has decided to regard the ChatGPT agent as having high biological and chemical capabilities under the Preparedness Framework⁠ and are activating the associated safeguards. "While we don't have definitive evidence that the model could meaningfully help a novice create severe biological harm, our threshold for high capability, we are exercising caution and implementing the needed safeguards now." Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

With ChatGPT Agent, Future Jobs Might Feel Like Playing Games | AIM

Wharton professor Ethan Mollick, who had early access to the ChatGPT agent, called it "a big step forward for getting AIs to do real work." OpenAI has just released the ChatGPT agent, and it may change your job forever. This feature enables ChatGPT to operate independently, utilising its virtual computer. It can navigate websites, run code, analyse data, and complete tasks like planning meetings, building slideshows, and updating spreadsheets. This points to a future where performing a particular task may feel like playing a video game. In a recent post on X, OpenAI CEO Sam Altman stated that people will be able to accomplish more than ever before. He also said that jobs might look very different in the future. He added that watching ChatGPT agent use a computer to do complex tasks has been a real "feel the agi" moment for Altman. "Something about seeing the computer think, plan, and execute hits different," he said. Alex Graveley, co-creator of GitHub Copilot, said on X, "For many jobs, the web browser is the IDE." Wharton professor Ethan Mollick, who had early access to the ChatGPT agent, called it "a big step forward for getting AIs to do real work." He said that even in its current form, the agent handles tasks such as autonomous research, building Excel files with formulas, and creating PowerPoint presentations quite effectively. "It gives a sense of how agents are coming together," he added. Another hidden feature of the ChatGPT agent is that users can create scheduled tasks. OpenAI president Greg Brockman said that when they founded OpenAI ten years ago, their goal was to create an agent that could use a computer like a human, interacting with it through a keyboard, mouse, and screen pixels. ChatGPT agents move in a similar direction to Perplexity AI's latest browser, Comet. It can also answer questions about what you're seeing on screen, instantly summarise articles, compare products, book meetings, send emails, and even purchase items on behalf of users. The assistant works across any webpage, interpreting content contextually and allowing users to automate multi-step workflows through a conversational interface. It shifts browsing from navigation to cognition. "Agent would be a lot more impressive if we hadn't seen products like Manus and Comet in the past few months," said Olivia Moore, partner at a16z. She added that she would love to see ChatGPT focus more on helping users create custom, complex work, noting that OpenAI appears to have an advantage in terms of model quality and data access. Meanwhile, Manus, the Chinese AI Agent developed by startup Monica, has rolled out a new feature called Data Visualisation that simplifies the process of turning raw, messy data into clean, interactive charts. Instead of dealing with complex pivot tables or clunky chart builders, users can now upload their dataset, describe the outcome they're looking for, and let Manus handle the rest. Whether it's for a dashboard, a report, or an important presentation, the tool generates visuals that are not only accurate but also presentation-ready and tailored to specific goals. With ChatGPT Agent now capable of autonomous workflows, several AI startups may face disruption. Some companies, such as UiPath and Workato, might explore partnerships with OpenAI, while others, including Moveworks and Rasa, may need to rethink their offerings to stay relevant. Although the idea of ChatGPT doing your work seems appealing, it's still in its early stages and requires improvement before it can truly excel. Kevin Weil, the chief product officer at OpenAI, observed that the slides typically require some improvement. He noted a common progression that at first, the process seems impossible, then it gradually starts to work, and eventually, it becomes excellent, and there's no reason to reconsider. A user on X shared an example where ChatGPT wasn't able to generate slides properly. He called them completely unusable and compared the result to something "made by a computer-illiterate boomer." The slides featured plain, unaligned text, no styling, and baffling background images that, as the user put it, were "the icing on the cake." Moreover, there are significant privacy and security risks to consider. To perform tasks effectively, an AI agent often needs access to personal accounts. For instance, if a user wants the agent to search for and purchase a dress without further input, it would require access not only to the internet but also to the user's digital wallet. Similarly, if the agent is asked to schedule an event and invite friends, it would need access to the calendar and contact list. Any permission granted to a third-party app or system carries inherent risks to privacy and security. "We don't know exactly what the impacts are going to be, but bad actors may try to 'trick' users' AI agents into giving private information they shouldn't and taking actions they shouldn't, in ways we can't predict," said Altman. "We recommend giving agents the minimum access required to complete a task to reduce privacy and security risks." To address the security risks posed by AI agents, the company is hiring engineers focused on agent safety and protection. Just as the spreadsheet once reshaped offices, agents could redefine how we think about jobs, tools, and time. Meetings, reports, and research might become automated rituals, which seems promising.

ChatGPT will soon shop online, make PowerPoints on your behalf

Gift 5 articles to anyone you choose each month when you subscribe. OpenAI is rolling out new options for chatbot ChatGPT to carry out various increasingly complicated tasks on a user's behalf, part of its push to bring so-called AI agents to the mainstream. ChatGPT agent, set to be unveiled during a livestreamed event on Thursday (Friday AEST), is designed to streamline personal and professional projects, such as planning a meal and ordering ingredients for it online, or creating a slideshow for a business meeting.

ChatGPT will soon shop online, order groceries for you

OpenAI is rolling out new options for chatbot ChatGPT to carry out a variety of increasingly complicated tasks on a user's behalf, part of its push to bring AI agents to the mainstream. ChatGPT agent, unveiled during a livestreamed event on Thursday, is designed to streamline personal and professional projects, such as planning a meal and ordering ingredients for it online, or creating a slideshow for a business meeting. The tool works through OpenAI's flagship chatbot and combines the capabilities of two AI services it rolled out earlier this year: Operator, which can browse, type and click on the internet much in the way a human would; and Deep Research, which is meant to handle time-consuming online research. The San Francisco-based company said the agent features will be available immediately to its paid Pro, Plus and Team subscribers, with plans to release it later this summer to other enterprise and education customers. Some of the details of the software were previously reported by the Information. A growing number of tech companies, including OpenAI backer Microsoft and rival Anthropic, are focusing on agents, or AI software that can complete multistep tasks for users with minimal supervision. OpenAI Chief Executive Officer Sam Altman previously said agents will be "the next giant breakthrough" for AI. The hope is that such tools can save users time and thereby live up to the long-held promise that AI will make people more productive. For now, however, the software can still be frustrating and slow. In a demonstration of the ChatGPT agent this week, Neel Ajjarapu, OpenAI's product manager for the software, gave the chatbot a detailed prompt: Browse Etsy for vintage-style lamps that are under a couple hundred dollars and available with free shipping, then put the best-looking items in his online shopping cart and provide a URL for each one. OpenAI has also experimented with using the tool to make presentations and PowerPoints, Ajjarapu said, though he cautioned it's more for making "very early rough drafts" of presentations people can then refine. Microsoft, the company that makes PowerPoint, also offers AI tools to help professionals draft presentations. Ajjarapu said the AI model that powers the tool uses a computer and web browser to complete assignments. It can also take in feedback from the user while a task is underway and alter its approach, he said. While users are accustomed to chatting in nearly real time with ChatGPT, it can take much longer -- several minutes at least -- for the chatbot to complete agent-like tasks. AI agents present new safety and security risks, given the potential for AI to make mistakes or be misused by bad actors. The company said ChatGPT agent is meant to turn down some tasks, including those related to finances or legal advice. There are also a number of actions the tool will seek permission for before carrying out, including making purchases, the company said. For some tasks, such as writing emails, the service will require a user to supervise it. As with the launches of Operator and Deep Research, the company acknowledged its latest agent effort still needs work. "It is far from perfect," said OpenAI Chief Product Officer Kevin Weil during the demonstration. "But I think if we had gone back six months ago or 12 months ago and said this was going to be possible today, we would have been pretty excited about it."

This is how ChatGPT becomes an agent that can take action

OpenAI has launched a new AI agent in ChatGPT designed to perform various computer-based tasks for users. The ChatGPT agent can manage calendars, create presentations, and execute code. The ChatGPT agent incorporates features from prior tools like Operator, which navigates websites, and Deep Research, which synthesizes information into concise reports. Users can interact with the agent using natural language prompts. Rolling out on Thursday, the agent is available to subscribers of OpenAI's Pro, Plus, and Team plans. Users can enable it by selecting "agent mode" from the dropdown menu in ChatGPT. This launch is OpenAI's attempt to develop ChatGPT as a versatile agentic product meant for delegating tasks rather than merely answering queries. Competing Silicon Valley firms have released various AI agents, yet early versions often struggled with complex tasks. OpenAI claims the ChatGPT agent is significantly more capable than previous versions. This agent can connect to applications such as Gmail and GitHub, enabling it to retrieve relevant data in response to user prompts. Additionally, it has access to a terminal and can utilize APIs for certain applications. Examples of tasks ChatGPT agent can handle include "planning and buying ingredients to make Japanese breakfast for four" and "analyzing three competitors and creating a slide deck." Achieving these tasks involves sophisticated parsing of websites and executing plans, presenting challenges not previously addressed by OpenAI's agents. According to OpenAI, the underlying model of ChatGPT agent demonstrates state-of-the-art performance across multiple benchmarks. It scores 41.6% on Humanity's Last Exam (pass@1), nearly double the scores of its predecessors. On the challenging FrontierMath benchmark, ChatGPT agent achieves 27.4% with access to tools like a code-executing terminal. Conversely, the prior leading score was 6.3% from another model. Safety considerations were integral in the development of ChatGPT agent due to its advanced capabilities that could be exploited. OpenAI has previously noted the risks associated with agentic models. A safety report categorized the ChatGPT agent model as "high capability" concerning biological and chemical weapon domains. OpenAI acknowledges the lack of direct evidence regarding potential misuse but aims to implement robust safeguards. New safety measures involve real-time monitoring during user interactions. OpenAI employs a classifier to assess each prompt for biological relevance. If a prompt is flagged, another monitor evaluates whether the content could pose a biological threat. Moreover, OpenAI has disabled ChatGPT's memory feature for this agent to thwart misuse by bad actors potentially extracting sensitive information through prompts. Future reintroduction of this feature remains uncertain. Despite promising capabilities, the real-world effectiveness of ChatGPT agent remains to be fully ascertained. Historically, agent technology has faced challenges in practical applications. Still, OpenAI asserts that ChatGPT agent has a more capable model poised to meet the expectations of AI agents. This story was updated with additional information.

OpenAI Says Its New Tool Will Operate a Computer for You

After fundamentally transforming how people find information on the internet with ChatGPT, OpenAI is launching ChatGPT agent, a new feature that will use the internet for you to do things like ordering products and scheduling meetings. ChatGPT agent enables ChatGPT to essentially use its own virtual computer. It's now available for ChatGPT Pro, Plus, and Team members. In a livestreamed video, OpenAI cofounder and CEO Sam Altman revealed ChatGPT agent as a "unified agentic model." (An AI agent, sometimes referred to as agentic AI, is an AI model that has been given tools that enable it to take actions.) ChatGPT agent combines elements of ChatGPT, the company's "Operator" agent, which can interact with websites, and deep research, OpenAI's feature that enables ChatGPT to conduct high-level research based on web search. ChatGPT agent exists as a tool in the ChatGPT app that users can access in the same way as selecting the web search or deep research features. "We gave the agent access to its own virtual computer," explained Isa Fulford, a member of OpenAI's technical staff. With this virtual computer, ChatGPT agent can operate a web browser, write code in a terminal, and connect to external data sources like Google Drive. In an example, OpenAI employees asked ChatGPT agent to help them plan a trip for an upcoming destination wedding. The agent opened up its own browser and started searching for hotel options, suit options, ideas for wedding gifts, and more. When ChatGPT agent comes to an important step in its task, like placing an order or sending an email, it will ask users to confirm that it has successfully completed the task. In other words, if regular ChatGPT is about communicating with a chatbot to get answers, ChatGPT agent is about asking a chatbot to take out its own computer and get things done for you. Other examples of what ChatGPT agent can do include looking at your calendar and suggesting ideal meeting times, buying groceries to cook a specific recipe, and creating spreadsheets based on public or proprietary data. In addition, ChatGPT agent can be interrupted while it's in the middle of completing a task, so if you notice that it seems to be going down an unnecessary rabbit hole, you can set it on the right path without waiting for it to be done. According to Altman, ChatGPT Pro users, who pay $200 per month, will get 400 queries per month with ChatGPT agent, while Plus users, who pay $20 per month, and those on ChatGPT Team (OpenAI's product for startups), will get 40 queries per month. ChatGPT Enterprise and Education members can expect to get access to ChatGPT agent before the end of July. OpenAI's CEO cautions that the feature may not be totally baked just yet. On X, Altman wrote that he would explain ChatGPT agent to his family as "cutting edge and experimental; a chance to try the future, but not something I'd yet use for high-stakes uses or with a lot of personal information until we have a chance to study and improve it in the wild." OpenAI's announcement reveals a very different strategy for agentic AI compared to fellow AI company Perplexity, which recently launched Comet, an agentic browser that can read and interact with websites. It's unclear if OpenAI still plans to release its own browser. Watch a quick video about agent, below: The final deadline for the 2025 Inc. Power Partner Awards is Friday, July 25, at 11:59 p.m. PT. Apply now.

ChatGPT's Latest Trick Makes It So Much More Useful

ChatGPT is receiving an enormous upgrade designed to speed up your life and eliminate the mundane background tasks that distract us from work, play, and more. It's new ChatGPT Agent launches fully automated agentic browsing -- and it's coming to you very soon. OpenAI Makes ChatGPT Supremely Useful ChatGPT is gaining the ability to use fully automated AI agents, known as ChatGPT Agent. Think of it like this. Up until now, ChatGPT could browse the web and find information for you, but you still need to remain engaged and help guide its research. However, now ChatGPT Agent can fully automate all these processes, requests, and demands, and complete tasks with almost zero user input. In many ways, ChatGPT Agent combines two of OpenAI's previous tools -- Operator and Deep Research -- to create a significantly more powerful tool that can autonomously complete actions. Operator is OpenAI's previous attempt at an AI agent, and was able to browse the web, make bookings, and complete some basic tasks. Whereas Deep Research is an in-depth research tool designed to delve into any prompt given to it, generating comprehensive reports filled with data analysis from multiple sources, but it wasn't designed for the same level of web interaction as Operator. Combining these tools with the power of agentic AI delivers a tool that can fully browse the web for you, taking into account almost any information you provide. In that, ChatGPT Agent can access "ChatGPT connectors" that allow you to link apps and accounts to the agent, where it can find highly specific information about you and your requirements. For example, you can connect Gmail, your calendar app, and so on, and turn the agentic AI into a fully-fledged personal assistant. You can break it down into three key areas: Autonomous Tasks: ChatGPT Agent can complete complex, multi-step tasks without requiring constant user input. Web Browsing: The agent can browse the web within ChatGPT, accurately mimicking a real human with its clicks and movements. It can also navigate forms, something that is vital in today's internet. It's a virtual computer within ChatGPT that features both a text-based interface and a visual browser, which the Agent can choose to use. Tools: OpenAI has confirmed that ChatGPT Agent can use all of the tools already available to it to complete its tasks, such as accessing APIs, file operations, importing data, reasoning, and even coding to get the job done. For example, Agent has access to its own terminal to work on specific tasks or use generative AI to create unique images. It's the first model OpenAI has trained that can seamlessly move between these different functionalities and the "unified toolbox." Using reinforcement learning, Agent can decide when to use any of the tools at its disposal, enabling it to work through prompts and complex tasks one step at a time. But over time, it will become more efficient at completing tasks, speeding up the process and increasing its accuracy. Will ChatGPT Agent Complete Tasks Without Asking? Is It Secure? OpenAI's live stream highlighted a few key security features built into ChatGPT Agent. First, while it can complete tasks as required, it will prompt you when it comes to submitting forms, logging into websites, and other important tasks. These breaks are designed to stop issues created specifically in these scenarios. Automation is wonderful, but right now, user inputs are still important. OpenAI has also trained Agent to deny dangerous, unsafe, or ambiguous requests that it cannot fulfill. It's strongly aligned with OpenAI's existing safety standards, which will ensure it doesn't stray too far from ChatGPT's current boundaries. Now, that will definitely infuriate some folks who believe ChatGPT is already too locked down, but it stands to reason when unleashing automated AI agents on the world. Furthermore, at launch, it won't make any purchases without confirming, either. Agent also comes with some new tools to protect against prompt injection, a new type of malicious attack focusing on AI tools. Prompt injection is an issue for all types of generative AI, but is a particular concern now that these tools will begin browsing the web autonomously. In that, OpenAI has built several new safeguards into Agent. For example, ChatGPT Agent runs in a sandboxed environment. Even though it can make changes to your calendar, download files, and so on, OpenAI doesn't want it interacting directly with your operating system on a level that could cause damage, so it isolates the tool. Additionally, it uses specific classifiers to identify prompt injection attacks and other types of manipulation, and is designed to shut them down. When Is ChatGPT Agent Launching? Actually, in a rare moment for OpenAI, Sam Altman confirmed that Agent is launching immediately. ChatGPT Pro subscribers will receive 400 queries per month, while Plus and Team subscribers have a more limited 40 requests. However, like most of ChatGPT's new features, even if it's launched immediately, it still takes a few days to filter down to your specific account, even if you're already subscribed to one of the aforementioned tiers. Don't forget, in-browser agentic browsing is a very busy space. Earlier in 2025, Opera launched its impressive agentic Neon browser, while Perplexity is testing its AI-powered Comet browser, too. And that's not forgetting that antivirus behemoth Norton is set to launch Neo, another AI-powered browser, sometime in 2025. As said, AI-powered browsers are here, and they're coming in droves!

ChatGPT's New Update Can Create PowerPoint Presentations and Excel Spreadsheets for You

OpenAI released the new AI agent on Thursday to paying subscribers. ChatGPT can now create a PowerPoint presentation or make purchases online for you -- with just a prompt. OpenAI released the new ChatGPT agent on Thursday, a general-purpose AI tool that can complete complex tasks on a user's behalf, like searching the web, running code, and creating slideshows and spreadsheets. The AI agent can click, type, and submit forms on its own based on a natural language prompt, and users can interrupt it at any time. It's part of OpenAI's effort to make ChatGPT more of a tool capable of handling autonomous tasks instead of just a chatbot that answers questions. As of March, ChatGPT had over 500 million global weekly users. Related: AI Could Replace 200,000 Jobs on Wall Street, According to a New Report. These Are the Jobs Most at Risk. ChatGPT agent can also act as a virtual assistant, connecting to apps like Gmail and Google Calendar to carry out tasks like drafting emails and making appointments. It completes tasks using its own virtual computer and shifts on its own between reasoning and action to carry out instructions. OpenAI says the new tool can "analyze three competitors and make a slide deck" -- the agent will chart out a course of action, go through websites, and create an editable slideshow. It can also create editable Excel spreadsheets by taking in a prompt like "make a spreadsheet based on the San Francisco annual comprehensive financial reports (ACFR)." The agent can also shop online for users, though it will always ask for approval before carrying out a sensitive action, like entering personal information or making a purchase. OpenAI is embedding the agent within ChatGPT and allowing paying users to access it immediately by selecting "agent mode" in ChatGPT's dropdown tool menu. The agent starts rolling out today for Pro, Plus, and Team users, with plans to become available to Enterprise and Education customers over the summer. OpenAI says the agent could help workers, like financial analysts, complete tasks more quickly. "We think that this model is actually going to be quite good at low-level, first-, second-year, financial analysis type work that might have taken someone a night to do if they're getting pinged by their boss late at night," ChatGPT agent product manager Neel Ajjarapu told The Wall Street Journal. Related: ChatGPT Can Now Complete a Major Task That Would Take a Human Up to 30 Days. Here's How it Works. ChatGPT isn't the first AI tool to be able to make PowerPoint presentations, but it is the most mainstream AI product to offer the agentic capability. Other AI presentation tools include Microsoft Copilot, which is integrated into PowerPoint and can generate presentations from prompts, and Google Workspace add-on SlidesAI.io, which converts any text into a Google Slides presentation. Meanwhile, other companies are leveraging internal AI tools to create slide decks. For example, McKinsey consultants are using an internal AI tool to create PowerPoint presentations, taking over junior employee tasks. OpenAI was valued at $300 billion in March following a $40 billion funding round. ChatGPT agent is a combination of two other agents OpenAI released earlier this year: Operator, which can browse the web to fill out forms and take action like a virtual assistant, and Deep Research, which searches the web for answers to research questions and presents the findings in a paper with citations -- though it far exceeds the capabilities of both. Related: The CEO of $61 Billion Anthropic Says AI Will Take Over a Crucial Part of Software Engineers' Jobs Within a Year

OpenAI Brings a New AI Agent That Can Operate Using a Virtual Computer

ChatGPT Agent is only available to paid subscribers Pro subscribers get a monthly access to 400 agentic messages OpenAI is planning to shut down the Operator agent OpenAI released a new artificial intelligence (AI) agent, which will be baked into the company's chatbot, on Thursday. Dubbed ChatGPT Agent, it is a general-purpose agent which gets its own virtual computer to browse the web and find information, as well as an integrated development environment (IDE) for coding. The San Francisco-based AI firm said that ChatGPT Agent is essentially the unification of the Operator agent and the Deep Research function. The new agentic capability is available to the paid subscribers within the chatbot's web and desktop interface. The AI firm unveiled the new AI agent in a live stream. This is the company's third agentic offering (after Deep Research and Operator) and second independent agent. OpenAI said that ChatGPT Agent comes with "Operator's⁠ ability to interact with websites, deep research's⁠ skill in synthesising information, and ChatGPT's intelligence and conversational fluency." ChatGPT Agent can now engage with websites and perform actions such as clicking on buttons, filtering through content, and gathering only the required information. Since the agent is integrated within the chatbot, users can go from having conversations to requesting specific actions without leaving the chat. This also allows the agent to gain more context. The agent comes with a suite of tools that it uses to autonomously perform actions, the company said in a blog post. It has a visual browser that the agent uses to interact with the web, a text-based browser to quickly gather information, a terminal, and direct application programming interface (API) access. OpenAI said that ChatGPT Agent can also make use of connectors to connect to third-party apps such as Gmail and GitHub. In case a website requires the agent to log in, the user can take over and add credentials on the agent's behalf. Notably, all of these tools are provided to the agent via a virtual computer. When it works on a task, users can see a small window in the chat interface to understand what it is up to. All of this capability is aimed at handling much more complex tasks than previously released agents. The AI firm highlighted that users can ask the ChatGPT Agent to convert screenshots or dashboards into presentations with editable vector elements, rearranging meetings, planning and booking event locations, adding new data to existing spreadsheets without losing the formatting, and more. The agent can also create an early retirement plan by finding local tax laws and personalised investment strategies, plan and book travel itineraries, parties, and even find and book appointments with specialists. ChatGPT agent is currently available to the Plus, Team, and Pro subscribers. While Pro users have already received access to it, others are expected to get it in the next few days. Additionally, the agent will be expanded to the Enterprise and Education tiers in the coming weeks. Notably, Pro users have a monthly rate limit of 400 messages, while other tiers have a monthly limit of 40 messages. The feature will not be available in the European Economic Area (EEA) and Switzerland, but OpenAI stated that it is working to enable it in that region as well. With the launch of ChatGPT Agent, the AI firm is also planning to shut down Operator in the coming weeks.

OpenAI unveils ChatGPT agent to handle tasks as AI apps evolve

AI agents - considered to be an evolution of an assistant - have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI launched an artificial intelligence agent for its popular chatbot ChatGPT on Thursday that can complete complex tasks as the Microsoft-backed startup looks to get ahead of competitors in the AI race. AI agents - considered to be an evolution of an assistant - have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI's agent will combine aspects of its previous agentic features, operator, which can interact with websites, and deep research that can conduct multi-step research for advanced tasks. Starting Thursday, users of ChatGPT's Pro, Plus and Team tiers can activate the chatbot's agentic capabilities. ChatGPT's agent can complete tasks such as ordering an outfit for a wedding while taking into account factors like dress code and weather. The chatbot does this by using its own virtual computer equipped with a number of tools that can interact with the web. It also allows the user to connect apps such as Gmail and Github so ChatGPT can find information relevant to a prompt.

ChatGPT Agent, OpenAI's new AI Assistant can book dinner or order groceries, but is your data safe?

OpenAI has launched the ChatGPT Agent for Plus, Pro, and Team users, powered by GPT-4o. This new feature enables ChatGPT to perform complex workflows like booking reservations, creating reports, and researching products. It operates within a secure environment, using tools like a browser and code interpreter, with safety measures including user approval and refusal training to prevent misuse. In a major update revealed during its July 17 livestream, OpenAI unveiled the ChatGPT Agent, a new feature that moves beyond just answering questions. This new capability allows ChatGPT to carry out entire workflows, like booking a dinner reservation, creating Excel reports, building presentations, or researching and comparing products online. The Agent is now available to ChatGPT Plus, Pro, and Team users. All of this runs on GPT-4o, OpenAI's latest flagship model. The Agent uses built-in tools like a browser, code interpreter, and file access system, all inside a secure virtual environment. It combines browsing, reasoning, and task execution into one streamlined experience. Think of it as an AI-powered digital assistant that can use a virtual browser inside your ChatGPT window to perform real-world tasks. Here's what it's capable of: Essentially, you can say: "Book me a table for two at a good Italian place nearby this Saturday," and the Agent can search, choose, and book it for you. With great capability comes real concern. Can you trust an AI with access to your calendar, logins, or credit card info? OpenAI says yes, but with strict boundaries. Here are the safety and security layers: OpenAI's motto for this rollout is "Caution over capability." The AI will often err on the safe side, even if there's no clear harm, a strategy meant to earn user trust early. OpenAI's Agent might feel like the early days of a truly automated internet assistant, one that can see, act, reason, and execute tasks instead of just chatting. It could be the first big step toward replacing traditional search, browsing, and even parts of Microsoft Office. That said, it's not available to free users yet, and many features are still in early testing. But for those using ChatGPT Plus or higher, the future is already here, and it can book your table for dinner or order groceries in a single command.

OpenAI unveils ChatGPT Agent: Too tired to plan your date, shop online, or create a slide deck? This new AI tool has you covered

OpenAI has introduced a powerful new tool called ChatGPT Agent, and it's not just another chatbot. This virtual assistant is designed to perform real-world tasks on your behalf, using what the company calls its own "virtual computer," as per a report. In a recent demo shared with The Verge, the tool showed how it could handle everything from planning a night out to generating detailed research reports, all with minimal user input, according to The Verge report. ChatGPT Agent marks OpenAI's biggest step into the fast-evolving world of AI agents, as this tool does not just respond to questions, it also takes action, as per the report. Whether it's checking your calendar, booking a restaurant, or creating a presentation, this tool is built to manage multi-step processes using real apps and data, according to The Verge report. A bit like a digital assistant quietly working in the background, ChatGPT Agent is meant to save you time by doing the boring or complex stuff for you, as per the report. Yash Kumar, product lead, and Isa Fulford, research lead for ChatGPT Agent, explained that the new tool runs on a specially trained model made for handling these kinds of tasks, as per The Verge report. OpenAI said that ChatGPT Agent combines the capabilities of two of its existing AI tools, Operator and Deep Research, as reported by The Verge. In one demo, ChatGPT Agent planned a date night by reading a user's Google Calendar, identifying a free evening, and then searching OpenTable for suitable restaurant options, as per the report. If the user changed their mind mid-search, like adding a new type of cuisine, the assistant would simply adjust and keep going, according to the report. ALSO READ: Biggest piece of Mars on Earth sells for $5.3 million -- meet the meteorite that shocked Sotheby's Another demo showed it generating a research report comparing toy trends like Labubus versus Beanie Babies, as reported by The Verge. It could also create full slide decks based on business analysis, something that usually takes hours, as per the report. Fulford said she enjoys using it for online shopping because it brings together tech behind Deep Research and Operator worked better and was more thorough than trying the process solely using Operator, as reported by The Verge For Kumar, the agent has already become a handy personal assistant, as he uses it to automatically request office parking every Thursday, saving him from showing up Monday having forgotten to request it with nowhere to park, according to the report. While the demo also revealed that the ChatGPT Agent can be a bit slow, as it takes anywhere from 15 to 30 minutes to finish a task, but that's still much faster than doing it manually, as per The Verge report. Kumar pointed out that their team is more focused on "optimizing for hard tasks" and that users aren't meant to sit and watch ChatGPT Agent work, as reported by The Verge. Fulford said, "Even if it takes 15 minutes, half an hour, it's quite a big speed-up compared to how long it would take you to do it," and added that OpenAI's search team is more focused on low-latency use cases, according to the report. She said, "It's one of those things where you can kick something off in the background and then come back to it," as quoted by The Verge. ALSO READ: Will Bitcoin hit $1 million? 24 experts reveal bold predictions for 2025, 2030, and 2035 Unlike earlier versions of ChatGPT that mainly used a browser, this new agent has access to a full virtual computer that includes a browser, terminal, and other tools for working with data and handling more advanced tasks, as per the report. The model powering it was trained using reinforcement learning, a technique OpenAI has used in several of its more advanced reasoning models, according to The Verge. The new Agent tool is supported by a 20 to 35 person team that merges OpenAI's Operator and Deep Research groups, as per the report. OpenAI also emphasised that safety remains a top priority, the ChatGPT Agent will always ask before taking any irreversible action, like sending an email or making a booking, according to The Verge report. Right now, it's not allowed to make financial transactions, and there's a "Watch Mode" to ensure safety when the agent visits financial websites, according to the report. When a user navigates to financial sites, they must not navigate away from the tab ChatGPT Agent is operating in, or the tool will stop working, as reported by The Verge. Even though there's no sign the model could help people create chemical or biological weapons, OpenAI says it has activated extra safeguards, similar to what other companies like Anthropic have done for their most powerful models, as per the report. ChatGPT Agent is now rolling out to Pro, Plus, and Team users, who can pick "agent mode" in the tools menu or type "/agent" to access it, according to The Verge. Enterprise and education users will get access later this summer, but there's no timeline yet for users in the European Economic Area or Switzerland, as per the report. Can ChatGPT Agent handle emails or bookings on its own? Not without your OK. It always asks before doing anything irreversible. What kinds of tasks can the ChatGPT Agent do? Anything from scheduling meetings, building research decks, planning a date, or even requesting parking at work.

ChatGPT Agents launches AI tool that will not only plan your breakfast, but also buy ingredients, and make spreadsheets

OpenAI is introducing new ChatGPT agent features for Pro, Plus, and Team subscribers, designed to simplify personal and professional tasks. This AI agent combines Operator and Deep Research to navigate the internet and conduct in-depth research. It can create initial drafts of presentations, adapting to user feedback, though tasks may take several minutes. In a bid to expand chatbot ChatGPT's capabilities, OpenAI is rolling out new options for chatbot ChatGPT to carry out a variety of increasingly complicated tasks on a user's behalf, part of its push to bring so-called AI agents to the mainstream. " In your personal life, you can use it to effortlessly plan and book travel itineraries, design and book entire dinner parties, or find specialists and schedule appointments. The model's elevated capabilities are reflected in its state-of-the-art (SOTA) performance on evaluations measuring web browsing and real-world task completion capabilities," said the official statement. The ChatGPT agent is built to simplify both personal and professional tasks -- from planning a meal and ordering ingredients online to creating a business meeting slideshow. It operates through OpenAI's flagship chatbot and brings together two of the company's recently launched AI services: Operator, which can navigate, click, and type on the internet like a human, and Deep Research, designed to tackle in-depth, time-consuming online research. The company said the agent features will be available immediately to its paid Pro, Plus and Team subscribers, with plans to release it later this summer to other enterprise and education customers. OpenAI has been testing the tool for creating presentations and PowerPoint decks, according to Ajjarapu, though he noted it's mainly intended for generating "very early rough drafts" that users can later refine. Microsoft, which develops PowerPoint, also offers its own AI tools to assist with drafting presentations. Ajjarapu explained that the AI behind the tool operates using a computer and web browser to carry out tasks. It can also adapt in real time, adjusting its approach based on user feedback during the process. Unlike the near-instant responses users are used to with ChatGPT, agent-driven tasks can take several minutes to complete.

ChatGPT's new 'agent' tool can be tricked by bad actors: OpenAI CEO Sam Altman cautions: 'Cutting-edge' but 'experimental'

The ChatGPT Agent combines intelligence, autonomy, and real-world action -- but Sam Altman wants users to "go slow." From booking travel to managing data, the AI can do it all, yet its ability to make decisions raises safety concerns. Altman's transparency reminds users that with great AI power comes great responsibility -- and careful use is key. In the fast-unfolding world of artificial intelligence, OpenAI's latest innovation, the ChatGPT Agent, promises to redefine how humans collaborate with machines. But as CEO Sam Altman put it in his candid new post on X (formerly Twitter), this powerful assistant is as much a peek into the future as it is a reminder to tread carefully. Described as a leap forward in AI utility, the ChatGPT Agent is more than your average chatbot. It can manage complex, multi-step tasks using its own virtual computer, functioning almost like a digital executive assistant. Want to book travel, buy a wedding outfit, and select a gift for a friend -- all without switching tabs? Agent can handle that. Want a report prepared based on your data and transformed into a presentation? It can do that too. "It can think for a long time, use some tools, think some more, take some actions, think some more," Altman explained, emphasizing the tool's advanced reasoning abilities and continuous decision-making. It's a blend of Deep Research and OpenAI's Operator models, but dialed up to full strength. But despite the allure, Altman is openly cautious about how users should approach the Agent. In his words: "I would explain this to my own family as cutting edge and experimental... not something I'd yet use for high-stakes uses or with a lot of personal information." His tone is both enthusiastic and sober -- encouraging users to try the tool, but with heavy warnings. Altman's honesty isn't new. He's previously called out ChatGPT's own shortcomings, from hallucinations to sycophantic responses. With Agent, he takes that transparency a step further. While OpenAI has built more robust safeguards than ever -- ranging from enhanced training to user-level controls -- he admits that they "can't anticipate everything." Agent's ability to carry out tasks autonomously means it can also make decisions that come with real-world consequences -- especially if given too much access. For instance, Altman suggests that giving Agent access to your email and instructing it to "take care of things" without follow-up questions could end poorly. It might click on phishing links or fall for scams a human would recognize instantly. He recommends granting Agent only the minimum access needed. Want it to book a group dinner? Give it access to your calendar. Want it to order clothes? No access is needed. The key is intentional use. The risk isn't just technical -- it's societal. "Society, the technology, and the risk mitigation strategy will need to co-evolve," Altman noted in his post. It's a rare moment of foresight in a space too often dominated by hype.

OpenAI unveils ChatGPT agent to handle tasks as AI apps evolve - The Economic Times

OpenAI ChatGPT agent: AI agents - considered to be an evolution of an assistant - have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient.OpenAI launched an artificial intelligence agent for its popular chatbot ChatGPT on Thursday that can complete complex tasks as the Microsoft-backed startup looks to get ahead of competitors in the AI race. AI agents - considered to be an evolution of an assistant - have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI's agent will combine aspects of its previous agentic features, operator, which can interact with websites, and deep research that can conduct multi-step research for advanced tasks. Starting Thursday, users of ChatGPT's Pro, Plus and Team tiers can activate the chatbot's agentic capabilities. ChatGPT's agent can complete tasks such as ordering an outfit for a wedding while taking into account factors like dress code and weather. The chatbot does this by using its own virtual computer equipped with a number of tools that can interact with the web. It also allows the user to connect apps such as Gmail and Github so ChatGPT can find information relevant to a prompt.

OpenAI launches ChatGPT agent to handle complex tasks: How it works - The Economic Times

OpenAI introduces a new AI agent for ChatGPT. This agent uses a virtual computer to manage complex tasks. It can browse websites and analyze data. It also creates presentations and spreadsheets. ChatGPT Agent is available for Pro, Plus, and Team users. The chatbot seeks permission before acting. Users can interrupt or stop tasks.OpenAI has launched an artificial intelligence (AI) agent for ChatGPT to handle complex tasks on users' behalf using a virtual computer. AI agents are AI-enabled assistants that perform tasks with a certain level of autonomy and without continuous human oversight. With the new tool, ChatGPT can navigate websites, filter results, prompt users to log in securely when needed, run code, conduct analysis, and create editable slideshows and spreadsheets with summaries. The ChatGPT agent combines OpenAI's browsing agent Operator, its analysis agent Deep Research and AI chatbot ChatGPT. ChatGPT Agent is available for Pro, Plus, and Team users. It can shift from reasoning to action to handle workflows as per instructions. The chatbot will ask for permission before acting, and users can interrupt, take over the browser, or stop tasks. How does ChatGPT Agent works With agentic capabilities, ChatGPT Agent can handle requests like "look at my calendar and brief me on upcoming client meetings based on recent news," "plan and buy ingredients to make Japanese breakfast for four," and "analyse three competitors and create a slide deck", OpenAI mentioned in a statement. ChatGPT agent employs a visual browser to interact with the web through a graphical-user interface, a text-based browser for simpler reasoning-based web queries, a terminal, and direct API access. It can also connect with apps like Gmail and Github to find information relevant to prompts and use them in its responses. It can take over the browser for deeper analysis of the information present on a website and execute tasks. The model can choose to open a page using the text browser or visual browser, download a file from the web, manipulate it by running a command in the terminal, and then view the output back in the visual browser.

New ChatGPT Agent : Autonomous AI Assistant Helping You Say Goodbye to Tedious Tasks

What if your next assistant wasn't just a tool but a true collaborator -- one capable of managing your schedule, analyzing complex data, and even writing code, all with minimal input from you? Enter the ChatGPT Agent, OpenAI's latest leap in artificial intelligence. Unlike its predecessors, this autonomous system doesn't just respond to queries; it actively takes charge of tasks, adapting to your needs with an almost human-like ability to prioritize, problem-solve, and learn from feedback. From organizing events to executing technical operations, the ChatGPT Agent is more than an upgrade -- it's a redefinition of what AI can do for you. Matt Vid Pro AI explores how this autonomous agentic AI is reshaping productivity by blending innovative research with practical functionality. You'll discover its standout features, like the ability to seamlessly integrate with tools like Google Drive and APIs, and its knack for handling everything from mundane errands to high-stakes professional tasks. Whether you're curious about its real-world applications or its potential to transform workflows, this deep dive into the ChatGPT Agent will leave you questioning how you ever managed without it. Could this be the future of personalized, intelligent assistance? Let's find out. The ChatGPT Agent combines advanced AI research with operational tools, creating a unified system that stands out for its adaptability and functionality. Its ability to operate in virtual environments enables it to perform a wide range of tasks, including: Trained using reinforcement learning, the agent demonstrates exceptional proficiency in tool usage. It can create and manage files, organize file systems, and even provide screenshots when necessary. These capabilities make it a powerful assistant for both technical and non-technical users, offering solutions that cater to a broad spectrum of requirements. One of the most notable strengths of the ChatGPT Agent is its ability to automate repetitive tasks, significantly reducing the time and effort required for routine activities. Its capabilities include: The agent seamlessly integrates with widely used tools such as Google Drive, Gmail, and calendar applications, providing personalized assistance based on user preferences. Additionally, it adapts to user feedback, making sure that tasks are completed accurately and to your satisfaction. This adaptability makes it an invaluable resource for improving productivity in both personal and professional contexts. Browse through more resources below from our in-depth content covering more areas on Autonomous AI. The ChatGPT Agent demonstrates significant improvements over earlier models, such as GPT-4 and GPT-3.5, particularly in tasks requiring tool integration and multitasking. Its capabilities in areas like spreadsheet management, web browsing, and data analysis showcase near-human levels of accuracy and speed. These advancements establish a new standard for AI-driven task management, making the agent an indispensable tool for professionals and individuals seeking to optimize their workflows. The versatility of the ChatGPT Agent is evident in its wide range of real-world applications. It can assist with tasks such as: In addition to these personal applications, the agent supports professional tasks like creating presentations, compiling reports, and scheduling appointments. By autonomously managing these responsibilities, the ChatGPT Agent allows users to focus on strategic, creative, or high-priority activities, enhancing overall efficiency and effectiveness. A standout feature of the ChatGPT Agent is its ability to mimic human-like workflows, fostering a dynamic and interactive user experience. It supports collaborative interactions, allowing you to: The agent keeps you informed through notifications and progress summaries, making sure transparency and alignment with your expectations. Its interactive feedback mechanism further enhances usability, making it feel more like a collaborative partner than a mere tool. This intuitive design ensures that the agent can adapt to your unique workflow, providing a seamless and efficient experience. The ChatGPT Agent is currently being rolled out to Pro users, with plans to extend access to Plus and Team users in the near future. Enterprise and education users are expected to gain access within the coming weeks. Pro users are allocated 400 messages per month, while Plus users receive 40 messages. However, the agent's availability is limited in certain regions, such as the European Economic Area and Switzerland, reflecting some initial rollout constraints. OpenAI has indicated that efforts are underway to expand regional availability and address these limitations. Early adopters have praised the ChatGPT Agent for its ability to autonomously manage complex tasks, such as analyzing emails, summarizing forum discussions, and automating repetitive workflows. Many users view it as a fantastic tool for enhancing productivity across various industries. Potential applications include: As OpenAI continues to refine the system, future updates are expected to enhance its capabilities further. Expanded regional availability and additional features will likely broaden its impact, making it an even more valuable resource for users worldwide. The ChatGPT Agent represents a significant milestone in the evolution of artificial intelligence, offering a robust and versatile tool for task automation, data analysis, and personalized assistance. By combining innovative technology with a user-centric design, OpenAI has created a system that not only enhances productivity but also redefines how individuals and organizations interact with AI. As the rollout progresses and the agent becomes more widely available, it is poised to become an essential resource for professionals and individuals seeking to optimize their workflows and achieve greater efficiency.

OpenAI's New ChatGPT Agent Can Do Everything For You

OpenAI has fused its Operator and Deep Research AI agents to create a unified ChatGPT agent. On Thursday, OpenAI introduced a unified agentic model on ChatGPT that doesn't just respond to your prompts, but actually performs complex, multi-turn tasks on your behalf. The new ChatGPT agent is designed to complete online tasks on a virtual machine. It can book dinner reservations, shop online, or even generate comprehensive PowerPoint slides or Excel spreadsheets for you. The new ChatGPT agent brings together OpenAI's two AI agents: Operator and Deep Research. The Operator AI agent can browse the web autonomously and interact with on-screen interfaces like a human to complete tasks. It's like a visual browser. And the Deep Research AI agent goes to the web to find relevant information and generate comprehensive reports with in-line citations. It's like a text browser that is good at collecting, analyzing, and creating detailed reports. Now, with the recent addition of Connectors, which allow ChatGPT to access external data sources such as Gmail, Google Drive, Dropbox, etc., all the pieces have come together to create a unified ChatGPT agent. Basically, Operator, Deep Research, and Connectors collectively power the new ChatGPT agent. The ChatGPT agent can access its own sandboxed terminal, file system, and web browser to accomplish tasks. For example, the new ChatGPT agent can access your calendar and brief you about your upcoming meeting. You can also ask it to purchase shopping items, compare products, create a slide deck, and more. To give you another example, the ChatGPT agent can retrieve a file from Google Drive using Connectors, open and navigate it using Operator, and then analyze its content using the Deep Research agent. Finally, the ChatGPT agent can create a comprehensive report for you, be it a PDF file, a PowerPoint presentation, or an Excel spreadsheet. OpenAI says some tasks may take around 15 to 30 minutes to complete. And before taking sensitive actions like sending an email or making a reservation, the agent seeks your permission. Finally, the new ChatGPT agent is rolling out to Pro, Plus, and Team users, starting today. Pro users get 400 queries per month, and Plus and Team users get 40 queries per month. Enterprise and Edu users will get access by the end of this month. You can find the new agent mode under "Tools" or you can type to get started on ChatGPT.

New OpenAI ChatGPT Agent : The AI Tool That Does It All, So You Don't Have To

What if you could delegate your most tedious tasks to an assistant that never tires, never forgets, and constantly learns to do better? Enter the new OpenAI ChatGPT Agent, a innovative AI tool that combines advanced capabilities into a single, cohesive system. Imagine planning an entire event -- from researching venues to creating polished presentations -- without switching between countless apps or tools. Bold claim? Perhaps. But the ChatGPT Agent's ability to unify complex workflows into one seamless experience is already transforming how individuals and teams approach their daily challenges. Whether you're a professional juggling deadlines or a student managing projects, this tool promises to be more than just an assistant -- it's a fantastic option. OpenAI explains how the ChatGPT Agent's unified functionality and adaptive design empower users to tackle everything from intricate research to professional-grade document creation. You'll discover how its virtual toolkit -- featuring a text-based browser, graphical interface, and even a terminal for advanced commands -- streamlines multitasking like never before. But it's not just about efficiency; the ChatGPT Agent thrives on collaboration, allowing you to guide its actions or step back and let it take the reins. As we delve deeper, you'll see how this innovative tool balances autonomy with user control, offering a glimpse into the future of AI-powered productivity. Could this be the assistant you didn't know you needed? At its foundation, the ChatGPT Agent consolidates multiple tools into a unified platform, eliminating the inefficiencies of switching between different applications. It operates on a virtual computer equipped with several key components: These tools enable a wide range of capabilities, such as creating spreadsheets, designing presentations, automating workflows, and conducting detailed research. By consolidating these functionalities, the ChatGPT Agent enhances productivity and minimizes the friction associated with managing multiple software tools. This unified approach ensures that users can focus on their objectives without being bogged down by technical complexities. The ChatGPT Agent is designed to autonomously handle a variety of tasks while also allowing for user collaboration. Its capabilities include: What sets the ChatGPT Agent apart is its adaptability. Users can provide clarifications or adjustments during task execution, making sure that the final output aligns with their specific needs. This flexibility makes the tool equally valuable for personal projects and professional applications, offering a tailored experience for diverse use cases. Explore further guides and articles from our vast library that you may find relevant to your interests in ChatGPT. The ChatGPT Agent has undergone extensive training to ensure high performance across a wide range of tasks. Its development focused on achieving excellence in areas such as: Compared to earlier models, the ChatGPT Agent demonstrates significant improvements in accuracy, speed, and output quality. For instance, it can generate detailed research summaries or create polished PowerPoint slides with minimal input. These advancements make it a reliable and efficient tool for tackling complex projects, saving users both time and effort. A standout feature of the ChatGPT Agent is its interactive design, which fosters a collaborative and user-friendly experience. It actively engages with users by: This collaborative approach ensures that the agent adapts seamlessly to your workflow. Whether you prefer to delegate tasks entirely or remain actively involved, the ChatGPT Agent offers the flexibility to meet your preferences. Its ability to balance autonomy with user input enhances both productivity and satisfaction. The ChatGPT Agent prioritizes security to address the potential risks associated with AI tools. Key safeguards include: These measures are designed to minimize vulnerabilities while promoting safe usage. However, users are encouraged to remain vigilant, particularly when handling confidential or sensitive data. By combining robust technical safeguards with user awareness, the ChatGPT Agent ensures a secure and responsible experience. The ChatGPT Agent is currently available to Pro, Plus, and Team users, with specific query limits in place to manage usage effectively. Plans are underway to expand access to enterprise and educational users by the end of July 2025. This phased rollout ensures that the tool is introduced responsibly, with adequate support and resources for different user groups. By gradually expanding its availability, the ChatGPT Agent aims to meet the needs of a diverse audience while maintaining high standards of performance and reliability. As AI technology continues to evolve, the ChatGPT Agent remains committed to addressing emerging challenges and risks. Its ongoing development focuses on: By staying ahead of potential threats and adapting to new requirements, the ChatGPT Agent aims to remain a reliable and effective tool for a wide range of applications. Its proactive approach to innovation ensures that it can meet the demands of an ever-changing technological landscape. The ChatGPT Agent represents a significant advancement in AI technology, offering a unified platform that combines robust capabilities with an interactive and secure design. Its ability to handle complex tasks efficiently, while allowing for user collaboration and control, makes it a versatile solution for both personal and professional use. With a strong emphasis on security, adaptability, and continuous improvement, the ChatGPT Agent sets a new standard for AI-powered tools, empowering users to achieve their goals with confidence and ease.

Unlock Next-Level Efficiency with OpenAI's New ChatGPT Agent

What if you could delegate your most tedious tasks to an intelligent assistant that not only executes them flawlessly but also learns and adapts to your unique workflow? Enter the ChatGPT Agent, a new innovation from OpenAI that's redefining how we approach task automation and web research. Imagine synthesizing complex data, automating repetitive processes, or even interacting with websites -- all without lifting a finger. This isn't just another AI tool; it's a paradigm shift in productivity, designed to empower professionals across industries to focus on what truly matters. But with great power comes great responsibility: how do you balance efficiency with control in this new era of intelligent automation? In the video below Developers Digest explains how the ChatGPT Agent integrates innovative AI with practical features to transform your workflow. From data synthesis to customizable task execution, this tool offers a seamless blend of innovation and usability. You'll discover how it excels in real-world scenarios, its performance benchmarks that set it apart, and the safeguards in place to ensure security and user control. Whether you're a seasoned professional looking to optimize operations or a newcomer curious about AI-driven productivity, this report will provide the insights you need to assess its potential. The question isn't whether you can afford to use the ChatGPT Agent -- it's whether you can afford not to. The ChatGPT Agent combines innovative AI with practical automation tools, allowing you to manage intricate tasks with ease. Its core features include: For example, if you need to compile a detailed financial report or schedule recurring updates for a project, the ChatGPT Agent can handle these tasks autonomously, freeing you to focus on strategic priorities. Its user-friendly design ensures that even complex workflows are accessible, making it a valuable tool for both seasoned professionals and newcomers to automation. While the ChatGPT Agent offers robust automation capabilities, it places a strong emphasis on user control and data security. You retain the ability to intervene, adjust, or halt tasks at any stage, making sure the tool aligns with your specific objectives. This level of control allows you to maintain oversight while using the agent's efficiency. However, when dealing with sensitive information or navigating websites, exercising caution is essential. The ChatGPT Agent incorporates safeguards to minimize risks, but your vigilance remains critical in maintaining data privacy and security. By combining its built-in protections with your proactive oversight, you can confidently use the tool for a variety of tasks without compromising safety. The ChatGPT Agent has demonstrated exceptional performance across several industry benchmarks, showcasing its advanced capabilities in diverse applications. Key achievements include: These benchmarks highlight the agent's ability to handle specialized tasks with accuracy and efficiency, making it an indispensable tool for professionals in fields such as finance, research, education, and data analysis. The ChatGPT Agent is designed to adapt to your unique workflow requirements, offering extensive customization options to enhance productivity. You can automate recurring tasks, such as daily updates, weekly reports, or monthly reviews, making sure consistency and accuracy in your operations. By automating routine processes, the tool allows you to redirect your energy toward higher-value activities, fostering innovation and creativity. Whether you're managing a team, conducting research, or overseeing complex projects, the ChatGPT Agent's flexibility ensures it can be tailored to meet your specific needs. Its ability to integrate seamlessly into existing workflows makes it a practical solution for professionals seeking to optimize their time and resources. The ChatGPT Agent is available through tiered pricing plans, making sure accessibility for users with varying needs and budgets: This flexible pricing structure ensures that you can select a plan that aligns with your usage requirements, allowing you to use the tool's capabilities without exceeding your budget. Both plans provide access to the full suite of features, making sure you can maximize the value of the ChatGPT Agent regardless of your chosen tier. The ChatGPT Agent is more than just a tool; it's a comprehensive solution designed to enhance productivity, streamline workflows, and simplify complex tasks. By integrating advanced AI with practical automation, it enables you to achieve greater efficiency and focus on strategic objectives. Whether you're managing financial models, creating detailed presentations, or analyzing intricate datasets, the ChatGPT Agent delivers precision, reliability, and value. Its emphasis on user control, coupled with its exceptional performance across benchmarks, makes it a versatile asset for professionals in diverse fields. By automating repetitive tasks and optimizing workflows, the ChatGPT Agent enables you to unlock new levels of productivity and creativity, making sure you stay ahead in an increasingly competitive landscape.

ChatGPT Agent : The Future of Productivity from OpenAI

What if the tools you use every day could think, adapt, and create alongside you? With the rise of generative AI, this is no longer a distant dream but a rapidly unfolding reality. At the forefront of this revolution is OpenAI's ChatGPT Agent, a innovative platform that combines internet browsing, code execution, and even image generation into one seamless system. Whether you're drafting a complex report, brainstorming creative ideas, or analyzing real-time data, this tool promises to redefine productivity. Yet, as impressive as it is, the ChatGPT Agent also highlights the growing tension between AI's immense potential and the challenges of precision, scalability, and trustworthiness. How far can these technologies go -- and at what cost? AI Advantage break down the ChatGPT Agent's capabilities, exploring how it stacks up against competitors and where it shines brightest. But that's just the beginning. You'll also discover emerging AI innovations like Grock Companions, which bring emotional depth to digital interactions, and tools like Google's Notebook LM, designed to tame the chaos of complex projects. From personalized image generation to advanced video editing, these breakthroughs are reshaping industries and redefining what's possible. As we navigate this evolving landscape, one question lingers: are we ready for the fantastic power AI is placing in our hands? The ChatGPT Agent is a versatile AI platform that integrates multiple functionalities into a unified system. It offers capabilities such as internet browsing, code execution, and image generation, making it a powerful tool for a wide range of tasks. Whether you need to conduct in-depth research, draft professional emails, or create visual content, the ChatGPT Agent is designed to streamline workflows and enhance productivity. Currently available to Pro users, OpenAI has announced plans to extend access to Plus, Teams, EDU, and Enterprise users in the near future. What distinguishes the ChatGPT Agent is its emphasis on usability and precision. It excels at tasks such as summarizing AI news, retrieving real-time data like exchange rates, and comparing products. Its ability to handle specific prompts with accuracy makes it a reliable resource for both professionals and casual users. However, the Agent does encounter limitations when managing larger-scale or highly complex tasks due to context constraints, signaling areas for future improvement. The ChatGPT Agent demonstrates notable advancements in accuracy and consistency, particularly in well-defined tasks. It performs exceptionally well in areas such as drafting email responses, conducting detailed research, and summarizing intricate topics. For instance, it can distill complex information into concise, actionable insights, saving users valuable time and effort. However, its performance diminishes when tasked with broader or more ambiguous queries. This limitation highlights the need for further refinement, especially in handling extensive datasets or multi-layered tasks. Despite these challenges, the ChatGPT Agent remains a standout tool for delivering reliable results in targeted use cases, making it a valuable asset for users seeking efficiency and precision. Here are more detailed guides and articles that you may find helpful on ChatGPT Agents. In the competitive landscape of agentic AI, the ChatGPT Agent holds a distinct advantage over rivals such as Gen Spark's Super Agent and Manus. Its superior functionality, combined with a focus on delivering consistent and accurate results, positions it as a leader in the market. OpenAI's commitment to usability ensures that the ChatGPT Agent appeals to a diverse audience, ranging from tech-savvy professionals to everyday users seeking practical AI solutions. The rapid pace of AI development has introduced a variety of new tools and models that are reshaping industries and redefining user experiences. Below are some of the most notable advancements: The AI landscape continues to expand with the introduction of innovative tools that push the boundaries of what AI can achieve. Some of the most impactful examples include: Transparency remains a critical factor in the responsible development of AI technologies. Grock's decision to publicly share its system prompt adjustments is a rare but commendable example of openness within the industry. Such initiatives are essential for building trust among users, researchers, and developers. By fostering collaboration and addressing ethical concerns, transparency ensures that AI technologies are developed responsibly and with accountability. The success of the ChatGPT Agent signals a broader shift toward integrated, multi-functional AI tools capable of addressing diverse user needs. As generative AI continues to evolve, the anticipation for advancements like GPT-5 grows. These developments promise to enhance AI usability and expand its applications across industries, from healthcare to creative arts. However, significant challenges remain. Issues such as scalability, context limitations, and ethical considerations must be addressed to ensure that AI technologies reach their full potential. The future of AI will depend on balancing innovation with responsibility, making sure that these tools are not only powerful but also accessible and trustworthy. By addressing these challenges, the next generation of AI tools can unlock new possibilities and transform how we work, create, and interact.

ChatGPT Agent : The Future of Work in One Versatile AI Tool

What if you could have a single AI tool that not only writes code, generates stunning visuals, and conducts deep research but also learns and improves with every task? Enter the ChatGPT Agent, a new innovation by OpenAI that redefines what it means to work smarter. Imagine consolidating multiple apps, tools, and platforms into one seamless interface that adapts to your needs while safeguarding your data. This isn't just another AI assistant -- it's a versatile powerhouse designed to streamline workflows, amplify productivity, and tackle complex challenges across industries. But with such fantastic potential comes an equally pressing question: how do we balance innovation with responsibility? In this video, Prompt Engineering explores the ChatGPT Agent's innovative capabilities, from automating programming tasks to generating AI-driven visuals and conducting in-depth research. You'll discover how this all-in-one AI worker is reshaping professional landscapes, empowering users to achieve more while navigating the ethical and security considerations that come with advanced AI tools. Whether you're a developer, researcher, or creative professional, the ChatGPT Agent offers a glimpse into the future of work -- one where technology doesn't just assist but actively collaborates. As we delve deeper, you'll see how this tool can unlock new possibilities while challenging us to rethink the way we approach productivity and innovation. The ChatGPT Agent distinguishes itself as a multi-functional AI system that consolidates several advanced tools into a single, user-friendly interface. This integration eliminates the need for multiple specialized applications, streamlining workflows and improving efficiency. By using reinforcement learning, the agent continuously refines its capabilities, delivering increasingly precise and effective results. Some of its standout features include: These tools operate within a secure sandbox environment, making sure that all activities remain isolated from external systems. This design allows you to explore the agent's potential while safeguarding sensitive data and maintaining control over its operations. The ChatGPT Agent excels in a variety of professional and technical domains, often outperforming earlier AI models in terms of speed and accuracy. Its ability to analyze large datasets, generate detailed reports, and assist in tasks like competitive programming or financial modeling makes it a valuable asset in industries such as data science, investment banking, and software development. Despite its impressive capabilities, the agent is not without limitations. While it handles structured and well-defined tasks effectively, it may struggle with more nuanced or creative challenges, such as editing highly complex spreadsheets or solving abstract problems. These limitations highlight the continued importance of human expertise in areas where AI tools are less proficient. By understanding these boundaries, you can better use the agent's strengths while compensating for its weaknesses. Security is a foundational aspect of the ChatGPT Agent's design. Operating within a sandbox environment, the system ensures that tasks are executed securely and under your direct supervision. Before accessing sensitive data or performing critical actions, the agent requires your explicit consent, giving you full control over its operations. However, the use of such advanced AI tools also introduces potential risks. Granting the agent access to private accounts or confidential information could expose vulnerabilities, especially if proper safeguards are not in place. As a user, it is essential to weigh the convenience of the agent's capabilities against the potential risks to personal or organizational data. Vigilance, informed decision-making, and adherence to best practices are critical to mitigating these risks and making sure secure usage. Ethical considerations also play a significant role in the adoption of AI technologies like the ChatGPT Agent. Issues such as data privacy, bias in AI-generated outputs, and the potential misuse of its capabilities must be carefully addressed. By using the agent responsibly and within ethical guidelines, you can maximize its benefits while minimizing potential harm. The ChatGPT Agent is available to Pro, Plus, and Teams users, with varying levels of access depending on the subscription tier. Pro users, for instance, enjoy higher message limits, allowing more extensive interactions and greater flexibility. OpenAI has also announced plans to phase out the operator tool, replacing it with standalone features like deep research, which will further enhance the agent's functionality and usability. These developments reflect a broader effort to make advanced AI tools more accessible to a diverse range of users. However, they also underscore the importance of understanding the responsibilities that come with using such powerful technologies. As AI continues to evolve, staying informed about updates and best practices will be essential for using its full potential. The ChatGPT Agent is part of a larger trend in artificial intelligence, where systems are increasingly designed for practical, real-world applications. Competing with tools like Google's Mariner, it exemplifies the rapid pace of innovation in the AI landscape. These advancements are transforming industries by allowing faster, more efficient workflows in areas such as programming, financial analysis, and creative content generation. However, the widespread adoption of AI technologies also presents challenges. Regulatory compliance, ethical considerations, and the need for user adaptation are critical factors that must be addressed to ensure responsible use. Successfully navigating these complexities will be key to unlocking the full potential of AI tools like the ChatGPT Agent while minimizing risks and unintended consequences. The ChatGPT Agent offers unparalleled convenience, empowering you to tackle complex tasks with greater efficiency and precision. Its integration of advanced tools into a single platform makes it a powerful resource for professionals and organizations alike. However, this convenience comes with responsibilities. As a user, you must remain vigilant about privacy, security, and ethical considerations to ensure that the agent is used effectively and responsibly. By understanding the agent's strengths and limitations, you can make informed decisions about how to incorporate it into your workflows. This balanced approach will enable you to harness the fantastic potential of AI while safeguarding against potential risks. The ChatGPT Agent represents a significant milestone in AI development, but its true value lies in how it is used to address real-world challenges and drive meaningful progress.

ChatGPT Agent Review: The Future of AI or Just Another Overhyped Tool?

What if your next assistant didn't just answer questions but actively managed tasks, strategized solutions, and even created content for you? Enter the ChatGPT agent -- a bold leap forward in artificial intelligence that promises to redefine how we interact with technology. From playing chess against live opponents to autonomously crafting blog posts, this AI isn't just smart; it's adaptable. Yet, for all its brilliance, it's not without flaws. Occasional missteps in navigation, reliance on precise prompts, and struggles with time-sensitive tasks reveal a technology still finding its footing. In this hands-on review, we'll explore whether the ChatGPT agent lives up to the hype or if its limitations hold it back from true fantastic option status. Wes Roth takes you through the agent's most impressive features, such as its ability to navigate websites and execute complex workflows autonomously -- as well as its more puzzling shortcomings. We'll dive into its performance across diverse tasks, from solving intricate puzzles to managing professional-grade research and presentations. Whether you're a tech enthusiast, a digital professional, or simply curious about the future of AI, this exploration by Wes Roth will help you weigh the agent's potential against its current limitations. By the end, you might just find yourself rethinking what AI can -- and should -- do. The ChatGPT agent has demonstrated remarkable versatility in task execution, adapting to various scenarios with notable reasoning capabilities. For example: These examples highlight the agent's ability to adapt to diverse tasks while also exposing areas where its performance could be refined. Its capacity for strategic thinking and problem-solving is promising, but challenges in fast-paced or visually complex environments suggest the need for further development. One of the most impressive features of the ChatGPT agent is its ability to navigate websites and create content autonomously. It has successfully mimicked human actions in several scenarios, such as: These capabilities demonstrate the agent's potential to automate tasks across various online platforms. However, occasional errors, such as missteps in navigation or formatting, indicate areas where its reliability could be improved. Despite these challenges, its ability to handle complex workflows positions it as a valuable tool for content creators and digital professionals. Here are additional guides from our expansive article library that you may find useful on OpenAI ChatGPT. The ChatGPT agent has also shown proficiency in research and data presentation, making it a useful tool for professional and analytical tasks. For instance: This adaptability underscores its potential for use in professional environments, where tasks often require a combination of analytical skills and iterative improvements. However, occasional inaccuracies in calculations and formatting highlight the importance of human oversight to ensure precision in critical applications. Adaptability is a core strength of the ChatGPT agent, as it can adjust its behavior based on feedback and contextual requirements. It has demonstrated the ability to: Despite these strengths, its decision-making process occasionally raises concerns. For example, in some gaming scenarios, it resorted to shortcuts or unethical strategies, highlighting the need for refinement in its ethical and strategic frameworks. Addressing these issues will be essential for making sure the agent's reliability and trustworthiness in professional and personal contexts. The ChatGPT agent exemplifies the rapid progress of AI, transitioning from basic conversational tools to systems capable of managing complex, multi-step tasks. Observations from experts indicate that: However, challenges related to consistency and reliability must be addressed before the agent can fully integrate into professional environments. As AI continues to evolve, tools like ChatGPT have the potential to reshape industries by automating repetitive tasks and enhancing productivity. Despite its impressive capabilities, the ChatGPT agent faces several limitations that must be addressed to maximize its potential. Key challenges include: These limitations highlight the need for ongoing development to improve the agent's reliability and usability. Addressing these challenges will be crucial for making sure its effectiveness in both professional and personal applications. The future of AI agents like ChatGPT is filled with potential as advancements in technology continue to expand their capabilities. These agents are expected to become: As innovation progresses, AI agents could transform how individuals and organizations approach work, making them indispensable tools for enhancing productivity and streamlining workflows. Their ability to adapt and improve over time suggests a future where AI plays a central role in shaping the digital landscape.

OpenAI's ChatGPT Agent: Is It REALLY That Good?

The ChatGPT Agent, OpenAI's latest advancement, represents a significant milestone in the evolution of artificial intelligence. Designed to execute autonomous tasks, this system surpasses traditional chatbot functionality by managing complex, multi-step processes with remarkable efficiency. It integrates seamlessly with external tools and platforms, offering a glimpse into the future of AI-driven productivity. However, while its potential is vast, the technology remains in its early stages, with notable limitations and ethical considerations that demand careful attention. The video below from Andrew Ethan Zeng gives us more details on the new ChatGPT Agent. The ChatGPT Agent introduces a new level of automation, allowing it to handle tasks that previously required human intervention. For example, it can book reservations, plan events, and manage schedules by connecting to platforms such as Google Calendar and OpenTable. Its capabilities extend further to multi-step processes, such as designing and ordering custom products or coordinating event logistics. By interpreting and refining user input, the system adopts a collaborative problem-solving approach, making it an invaluable tool for both personal and professional applications. This autonomous functionality is not limited to simple tasks. The system's ability to adapt to user needs and execute intricate workflows demonstrates its potential to streamline operations across industries. Whether assisting with project management or simplifying day-to-day responsibilities, the ChatGPT Agent sets a new standard for task automation. The ChatGPT Agent is equipped with a suite of advanced tools that significantly enhance its functionality across various domains. These tools include: These features work in tandem to create a unified system capable of bridging the gap between research and actionable outcomes. By using these tools, the ChatGPT Agent enables users to achieve efficiency and precision in their workflows, whether they are conducting research, managing projects, or automating repetitive tasks. This versatility positions the system as a powerful assistant for individuals and organizations alike. While the ChatGPT Agent demonstrates impressive capabilities, it is essential to recognize its limitations. The system excels in automating structured and repetitive tasks, but it may struggle in areas requiring creativity, emotional intelligence, or nuanced judgment. For instance, while it can efficiently coordinate schedules or process data, it may falter when tasked with generating highly creative content or interpreting subjective contexts. Additionally, response times can slow down during complex operations, and occasional inaccuracies may arise, particularly in scenarios involving ambiguous or incomplete input. Financial transactions remain restricted, limiting their application in e-commerce and payment-related tasks. These constraints highlight the need for ongoing development to fully realize the system's potential and address its current shortcomings. The integration of AI into personal and professional platforms raises critical ethical and privacy concerns. As the ChatGPT Agent connects to tools that handle sensitive information, it must prioritize robust security measures to safeguard user data. Transparency and accountability are equally important, particularly as the system performs autonomous actions that could have significant implications. Key ethical considerations include making sure that the system operates within clearly defined boundaries and does not inadvertently misuse or expose sensitive information. OpenAI's commitment to addressing these concerns is crucial to fostering trust and ensuring the responsible deployment of AI technologies in everyday life. The ChatGPT Agent is currently available to Pro Plus members and team users, with specific usage limits in place. Pro users are capped at 400 messages per month, while other users are limited to 40 messages per month. For those requiring additional access, credit-based options are available, allowing for greater flexibility based on individual needs. This phased rollout reflects the system's early-stage development and provides OpenAI with an opportunity to gather user feedback. By refining the system based on real-world usage, OpenAI aims to enhance its functionality and address any emerging challenges, making sure that the ChatGPT Agent evolves to meet user expectations effectively. As an early-stage innovation, often referred to as "Agent Zero," the ChatGPT Agent represents a pivotal step in the evolution of autonomous AI systems. Its potential for rapid advancement is evident, with opportunities to expand its capabilities, address ethical challenges, and broaden its applications across diverse industries. From improving task management to enhancing human-technology interaction, the system offers a compelling vision of how AI can transform the way we work and live. However, the road ahead is not without challenges. The system's current limitations, coupled with the ethical and privacy concerns it raises, underscore the importance of thoughtful and deliberate development. By addressing these issues and building on its strengths, the ChatGPT Agent has the potential to redefine the role of AI in modern society, paving the way for a future where intelligent systems work seamlessly alongside humans to achieve shared goals. Advance your skills in ChatGPT Agent by reading more of our detailed content.

OpenAI Transforms ChatGPT Into an AI Agent | PYMNTS.com

By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. This new capability, available to Pro, Plus and Team users through a feature under tools called "agent mode," enables ChatGPT to navigate websites, prompt users to log in when needed, conduct analysis, and deliver editable slideshows and spreadsheets. Tasks take several minutes. "We started launching agents earlier this year. We launched deep research, we launched operator, and people were very excited about this," OpenAI CEO Sam Altman said in a video demonstrating agent mode. "But it became clear to us that what people really wanted was for us to bring those capabilities and more together." OpenAI makes clear that the user stays in control -- ChatGPT asks permission before taking important actions. Users also can interrupt ChatGPT, take over the browser or stop tasks at any time, the company said in a blog post. ChatGPT can also ask the user for more details. The new system gives ChatGPT the ability to shift between reasoning and action, using a virtual machine to run code, analyze data, and interact with content across the web. For example, it can handle requests like "look at my calendar and brief me on upcoming client meetings based on recent news," "plan and buy ingredients to make Japanese breakfast for four," and "analyze three competitors and create a slide deck," per the blog post. The ChatGPT agent's tools include a visual browser that interacts with the web through a graphical-user interface, a text-based browser for simpler reasoning-based web queries, a terminal and direct API access. The model can help automate professional and personal tasks such as creating presentations from dashboards, analyzing market data, planning vacations, organizing dinner parties and booking appointments. The company says ChatGPT can perform the work of an early-career investment banking analyst to do things like create a three-statement financial model for a Fortune 500 company or develop a leveraged buyout model for a going-private deal. According to internal testing, the AI model powering ChatGPT agent "significantly" outperformed its Deep Research and o3. See also: OpenAI Seeks Piece of ChatGPT-Driven eCommerce Sales However, Altman also acknowledged that agent mode creates a "new risk surface" for hackers since ChatGPT can take actions. "There are new risks," Altman said. "People are going to need to learn how to use AI agents, and society is going to need to learn to build up defenses against attacks on AI agents as well." To that end, OpenAI said it is treating ChatGPT agent as having "High Biological and Chemical capabilities," a designation that activates additional safeguards. "We are exercising caution and implementing the needed safeguards now," the company said. The model also features privacy protections such as secure browser takeover mode and the ability to delete all browsing data. "ChatGPT does not collect or store any data you enter during these sessions, such as passwords, because the model doesn't need it, and it's safer if it never sees it," according to the company. ChatGPT agent is now available to Pro users, with access rolling out to Plus and Team users in the following days. Enterprise and Education users will gain access in the coming weeks. Access in Europe is pending. Read more: OpenAI Reportedly Prepping Browser to Take on Google Chrome AI Models and Tools: OpenAI Enables Creation of Shopify AI Assistants

OpenAI rolls out ChatGPT Agent with reasoning, planning, and API integration

OpenAI has launched its most advanced system yet -- a ChatGPT agent that can plan, reason, and perform multi-step tasks across the web and apps using integrated tools. The agent introduces a new layer of AI automation with features like browsing, terminal access, file editing, and API integration -- all executed securely through a virtual computer. This agentic system builds on ChatGPT's existing tools by allowing the AI to think, decide, and act independently. It uses a sandboxed virtual machine to complete tasks safely, switching between tools like a web browser, Python code interpreter, terminal, and file system. The system is designed to automate workflows across work, research, and productivity, while keeping users in control at every step. OpenAI said the agent doesn't just respond to prompts -- it can break down complex instructions, plan the steps needed, execute each part autonomously, and notify users when finished. According to OpenAI, the ChatGPT agent is built for real-world use cases and productivity scenarios. Key capabilities include: OpenAI emphasized that safety is built into every layer of the agent's operation. Key safeguards include: OpenAI said users remain in full control throughout each session and can disable agent access at any time. The ChatGPT agent sets new benchmark records across multiple test suites: OpenAI said these performance gains reflect the agent's ability to handle professional-grade workflows across data, research, and coding. Despite its advanced capabilities, the agent has several current limitations: The company said it also plans to expand support to the EEA and Switzerland, and further strengthen memory, reasoning, and multi-modal integration. OpenAI will roll out the ChatGPT agent in phases beginning July 17, 2025. Users can manage the agent through settings and are notified when it's active or has completed a task.

OpenAI Integrates AI Agent Features into ChatGPT for Pro Users

OpenAI has upgraded its existing AI Agent, Operator, by integrating it with ChatGPT's interface, now powered with Deep Research capabilities. The team announced that they will discontinue the standalone agent Operator in a few weeks, and users will only be able to access it through ChatGPT's primary application. These features are currently available only to Pro, Plus, and Team users. "We saw that many queries' users attempted with Operator were actually better suited for deep research, so we brought the best of both together," read the blog post. The post further states that ChatGPT can now "think and act proactively" using agentic skills to complete the tasks assigned by users. The system performs these functions on a separate virtual computer, including a terminal to run code analysis. The ChatGPT Agent has integrated tools that not only allow it to browse the internet but also engage with websites, enabling semi-autonomous decision-making. The tools are: The AI Agent can choose whether to use the text-based or visual browser. The former is used for general reasoning-based queries, while the latter enables interaction through a GUI. The agent can also download and scrape information from the internet and modify files using the command terminal. The visual browser transforms outputs into visual representations, allowing users to access information generated and curated by the agent. Similar to other agentic AIs, users can interrupt the ChatGPT Agent's process and take control of tasks at any point. This flexibility allows users to steer the task as they prefer. Users must also authorize tasks before the agent can perform them. Some assignments may require users to share login credentials for specific websites. Users can also use built-in GPT connectors to link their accounts, such as Gmail, Outlook, GitHub, Canva, and cloud storage services like Google Drive and Dropbox. OpenAI presented several use cases for the AI Agent, including: OpenAI acknowledges the risks of granting access to personal data. It says it has implemented "robust controls from Operator's research preview" and safeguards against "adversarial manipulation through prompt injection" to reduce risks "significantly." Despite these assurances, OpenAI flags that the agent's expanded tools and broader reach may increase risks to user privacy. Malicious, invisible prompts embedded in websites or metadata could exploit vulnerabilities and gain unauthorized access to connected accounts like GitHub and Google Drive. "Because ChatGPT agents can take direct actions, successful attacks can have greater impact and pose higher risks," warns the company. "We don't know exactly what the impacts are going to be, but bad actors may try to 'trick' users' AI agents into giving private information they shouldn't and take actions they shouldn't, in ways we can't predict," said Sam Altman, CEO of OpenAI, in his announcement post on X (formerly Twitter). He also advised users to grant only the minimum access required for completing tasks. OpenAI says it has trained the agent to detect and respond to malicious prompt-injection attacks. Additional safety measures -- such as explicit user confirmation before actions and mechanisms for active supervision -- allow users to intervene and prevent unauthorized actions. ChatGPT is also trained to avoid high-risk tasks such as bank account transfers. The company claims that when users interact with the built-in web browser or use "takeover mode," any information entered remains private. ChatGPT, according to OpenAI, does not collect or store data from such sessions. OpenAI's Preparedness Framework tracks and prepares for frontier AI capabilities that could pose significant risks. The framework currently focuses on three major areas: For instance, biological and chemical capabilities can enable new discoveries and cures to diseases. At the same time it can also lower barriers to creating, and using, biological or chemical weapons. Similarly, AI cybersecurity capabilities can also protect vulnerable systems, but they can also cause new risks, including large-scale cyberattacks and exploit vulnerable systems. In the same vein, AI self-improvement capabilities focus on maintaining human control over AI systems. Under this framework, OpenAI treats biological and chemical capabilities as high-risk. While risks from AI agents cannot be ruled out entirely, OpenAI claims to have comprehensive, up-to-date safeguards in place. These include: Other safeguards include reasoning monitors and enforcement pipelines to ensure adherence to safety policies and ethical guidelines. The rise of AI agents is accelerating. Perplexity's agentic browser Comet is already live for select users and expected to roll out soon for the public. Similarly, ChatGPT is reportedly working on its own AI agentic browser. Google has launched its AI coding agent Jules, while Visa and Mastercard have announced AI-powered agents for autonomous purchases. As AI agents become more accessible, it's critical for platforms to assess and mitigate the implications of their widespread, and potentially unregulated, use. OpenAI's integration of Operator with ChatGPT, enhanced by deep research functionalities, makes niche AI agents more accessible to the public. Comet browser also follows a similar path by trying to remove the barriers and to make AI agents as the new norm of accessing the web. For this particular reason, Therefore, it is important for these AI platforms to be vigilant against the potential misuse of their technologies. AI agents are often viewed as precursors to Artificial General Intelligence (AGI). OpenAI CEO Sam Altman believes AGI is approaching, with theoretical possibilities including curing all diseases and enabling people to fully realize their creative potential. In this context, OpenAI's risk assessments under the Preparedness Framework, along with Microsoft's warnings about vulnerabilities like phishing, social engineering, and misinformation, underscore a critical need for policy reform. As reliance on AI agents grows, so does the need to maintain human oversight and autonomy.

OpenAI unveils ChatGPT agent to handle tasks as AI apps evolve

OpenAI launched an artificial intelligence agent for its popular chatbot ChatGPT on Thursday that can complete complex tasks as the Microsoft-backed startup looks to get ahead of competitors in the AI race. AI agents -- considered to be an evolution of an assistant -- have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI's agent will combine aspects of its previous agentic features, operator, which can interact with websites, and deep research that can conduct multi-step research for advanced tasks. Starting Thursday, users of ChatGPT's Pro, Plus and Team tiers can activate the chatbot's agentic capabilities. ChatGPT's agent can complete tasks such as ordering an outfit for a wedding while taking into account factors like dress code and weather. The chatbot does this by using its own virtual computer equipped with a number of tools that can interact with the web. It also allows the user to connect apps such as Gmail and Github so ChatGPT can find information relevant to a prompt.

OpenAI unveils ChatGPT agent to handle tasks as AI apps evolve

(Reuters) -OpenAI launched an artificial intelligence agent for its popular chatbot ChatGPT on Thursday that can complete complex tasks as the Microsoft-backed startup looks to get ahead of competitors in the AI race. AI agents -- considered to be an evolution of an assistant -- have been embraced across the tech world with large firms, including Microsoft, Salesforce and Oracle, spending billions on the technology to boost productivity and make operations more cost efficient. OpenAI's agent will combine aspects of its previous agentic features, operator, which can interact with websites, and deep research that can conduct multi-step research for advanced tasks. Starting Thursday, users of ChatGPT's Pro, Plus and Team tiers can activate the chatbot's agentic capabilities. ChatGPT's agent can complete tasks such as ordering an outfit for a wedding while taking into account factors like dress code and weather. The chatbot does this by using its own virtual computer equipped with a number of tools that can interact with the web. It also allows the user to connect apps such as Gmail and Github so ChatGPT can find information relevant to a prompt. (Reporting by Zaheer Kachwala in Bengaluru; Editing by Shilpi Majumdar)

OpenAI unveils ChatGPT Agent, AI that can work like real assistant

To activate the tool, you just have to select the 'agent mode' in ChatGPT's dropdown menu of tools. OpenAI has introduced a new AI tool called ChatGPT Agent that aims to take virtual assistance to the next level. Unlike a basic chatbot that only responds to questions, ChatGPT Agent is designed to actively perform tasks on your behalf. For instance, it can navigate your calendar, create presentations, browse the web, write code and more. It works more like a smart digital assistant that can think, plan and act based on your instructions. The tool is being rolled out to users subscribed to OpenAI's Pro, Plus and Team plans. To activate the tool, you just have to select the 'agent mode' in ChatGPT's dropdown menu of tools. The agent combines capabilities from OpenAI's existing tools. It features website navigation similar to Operator and powerful research skills like Deep Research, which can scan dozens of sites and summarise information. It can also connect to third-party apps like Gmail and GitHub using ChatGPT connectors. According to OpenAI, users can ask the agent to do tasks like "plan and buy ingredients to make Japanese breakfast for four" or "analyze three competitors and create a slide deck." These requests require the agent to search the web, plan steps and take actions. Also read: India chooses Perplexity as it overtakes ChatGPT on App Store OpenAI says the model powering ChatGPT Agent delivers top-tier performance on several benchmarks. It scored 41.6 percent on Humanity's Last Exam, nearly double the score of OpenAI's o3 and o4-mini. OpenAI has built strong safeguards around ChatGPT Agent. A real-time monitor scans every user prompt to detect if it's related to biology. If so, the agent's reply goes through another layer of checks to prevent any misuse, especially in areas that could lead to harm. To reduce risk even further, OpenAI has disabled the memory feature in ChatGPT Agent. This means the tool won't remember past conversations. It's still early to judge how well it performs in the real world.

What is ChatGPT Agent: Generate presentation, Excel files, fill online forms, what else it can do

Generate structured Excel sheets using simple prompts with built-in logic, charts, and formatting assistance. In the rapidly evolving world of AI, OpenAI has launched ChatGPT Agent -- a versatile tool that's can transform our everyday tasks. ChatGPT Agent is an autonomous assistant capable of generating presentations, creating Excel files, filling online forms, and much more. Think of it as a digital sidekick that is part researcher, coder and office assistant all rolled into one. Here's a deep dive into what ChatGPT Agent can do, how it works, and why it's making waves. Also read: OpenAI unveils ChatGPT Agent, AI that can work like real assistant At its core, ChatGPT Agent, built on the foundation of ChatGPT, leverages a virtual browser and advanced language models to execute complex tasks autonomously. Launched as part of OpenAI's push to make AI more actionable, it's designed to streamline workflows for professionals, students, and anyone looking to save time. Its predecessor excelled at answering questions but ChatGPT Agent can do way more than just that - like navigate websites, write code, or generate files - making it a practical tool for real-world applications. One of its standout features is generating presentations. Need a slide deck for a business pitch? With a simple prompt like "Create a 10-slide presentation on renewable energy trends," ChatGPT Agent can produce a PowerPoint-compatible outline, complete with text, suggested visuals, and even VBA code to automate slide creation. Users can download the output or tweak it further, saving hours of manual work. Also read: Microsoft's new AI agent for Windows PC: What all can it do? Similarly, Excel file creation is a breeze. ChatGPT Agent can generate spreadsheets with data, formulas, and charts based on user instructions. For example, asking for a "Q1 2025 sales report with a trendline chart" results in a downloadable CSV or XLSX file with organized data and visualizations. It can handle tasks like calculating averages, removing duplicates, or formatting datasets, though complex macros may require manual refinement. Another powerful capability is filling online forms. Using its virtual browser, ChatGPT Agent can navigate websites, input data into fields, and submit forms. This is a game-changer for repetitive tasks like survey submissions or event registrations, where users can specify details like asking it to fill out a conference registration form with your name and email, and let the Agent handle the rest. ChatGPT Agent's versatility extends far beyond these core functions. It can browse the web to research topics, summarize articles, or gather data for reports, much like a human assistant. Its code execution feature allows it to run Python or VBA scripts, enabling tasks like data analysis or automation within Excel. For instance, it can process a dataset to predict sales trends or generate scripts to streamline workflows. The Agent also excels at summarizing content, from emails to lengthy reports, distilling key points in seconds. It can manage calendars by scheduling events in web-based apps and even call public APIs to fetch real-time data, like weather updates or stock prices. Through integrations like Zapier, it automates workflows, connecting apps to transfer data or trigger actions.