DeepSeek's Rapid Rise Marred by Major Data Breach and Security Concerns

DeepSeek: Why the hot new Chinese AI chatbot has big privacy and security problems

The Chinese artificial intelligence (AI) company DeepSeek has rattled the tech industry with the release of free, cheaply made AI models that compete with the best US products such as ChatGPT. Users are rushing to check out the new chatbot, sending DeepSeek's AI Assistant to the top of the iPhone and Android app charts in many countries. However, authorities have sounded a note of caution. US officials are examining the app's "national security implications." Australia's former cybersecurity minister said national security agencies will soon issue formal guidance for users. Why are governments and security experts so concerned? The main issue is the app is made in China and stores data there -- but that doesn't mean all the worry is just xenophobia. What information does DeepSeek record? DeepSeek does not appear to be spyware, in the sense it doesn't seem to be collecting data without your consent. However, like many online services, it clearly tells you it will record a lot of data about you and your behavior. Specifically, the company's privacy policy says it collects three categories of information. First, there is information you provide directly, such as your name and email address and any text you type in or files you upload. Next, there is automatically collected information, such as what kind of device you are using, your IP address, details of how you use the services, cookies, and payment information. Finally, there is information from other sources, such as Apple or Google login services, or third-party advertising and analytics companies. DeepSeek says it uses this information for a range of purposes: to provide services, enforce terms of use, communicate with users, and review and improve performance. The policy also contains a rather sweeping clause saying the company may use the information to "comply with our legal obligations, or as necessary to perform tasks in the public interest, or to protect the vital interests of our users and other people." DeepSeek also says it may share this information with third parties, including advertising and analytics companies as well as "law enforcement agencies, public authorities, copyright holders, or other third parties." DeepSeek will also keep the information "for as long as necessary" for a broad range of purposes. Again, this is all fairly standard practice for modern online services. Causes for concern Much of the cause for concern around DeepSeek comes from the fact the company is based in China, vulnerable to Chinese cyber criminals and subject to Chinese law. DeepSeek stores the information it collects "in secure servers located in the People's Republic of China." The company says it maintains "commercially reasonable technical, administrative, and physical security measures" to protect the information. However, we should keep in mind that China is one of the most cyber crime-prone countries in the world -- ranking third behind Russia and Ukraine in a 2024 study. So even if DeepSeek does not intentionally disclose information, there is still a considerable risk it will be accessed by nefarious actors. China is home to a sophisticated ecosystem of cyber crime organizations that often build detailed profiles of potential targets. Microsoft and others have accused the Chinese government of collaborating with cybercrime networks on cybercrime attacks. These organizations can use personal information to craft convincing targeted phishing attacks, which try to trick people into revealing more sensitive information such as bank details. If you are an experienced user who is familiar with online privacy and the capabilities of modern AI systems, go ahead -- but proceed with caution and be very wary about what information you share. And if you're less experienced -- if you're a casual user who is less internet-savvy -- my expert advice is to stay well away. DeepSeek won't give you much you can't get from other chatbots such as ChatGPT or Claude, and it might make your data vulnerable to Chinese cyber criminals and subject to Chinese law. DeepSeek also raises questions for governments. Efforts to prevent scams and cybercrime often focus on banks, telecommunications companies, and social media platforms -- but what about chatbots?

DeepSeek's database was wide open -- did hackers get in?

DeepSeek, the trending Chinese artificial intelligence (AI) startup, recently exposed one of its databases on the internet, potentially allowing unauthorized access to sensitive data. The exposed ClickHouse database provided full control over its operations, according to Wiz security researcher Gal Nagli. The exposure included over a million lines of log streams featuring chat history, secret keys, backend details, and other critical information, such as API secrets and operational metadata. Following notification attempts from the cloud security firm, DeepSeek has since fixed the security vulnerability. The database, which was accessible at oauth2callback.deepseek[.]com:9000 and dev.deepseek[.]com:9000, enabled unauthorized users to execute arbitrary SQL queries via the web browser without requiring authentication. It remains unclear if any malicious actors accessed or downloaded the data before the issue was resolved. Nagli emphasized the risks of rapid AI service adoption without adequate security measures, highlighting that real risks often stem from basic oversights like accidental database exposure. He stated, "Protecting customer data must remain the top priority for security teams, and it is crucial that security teams work closely with AI engineers to safeguard data and prevent exposure." DeepSeek has garnered significant attention for its innovative open-source AI models that aim to compete with established systems like OpenAI, positioning its reasoning model R1 as an "AI's Sputnik moment." Its AI chatbot rapidly climbed to the top of app store rankings across multiple markets, even as the company faced "large-scale malicious attacks," which led to a temporary pause in new registrations. In a January 29, 2025 update, DeepSeek acknowledged the database issue and indicated that it is implementing a fix. Concurrently, the company faces scrutiny regarding its privacy policies, with its Chinese affiliations raising national security concerns in the United States. In related developments, DeepSeek's applications became unavailable in Italy after the country's data protection regulator, the Garante, sought information regarding the startup's data handling practices and its sources of training data. The withdrawal of apps from the Italian market may or may not have been a direct response to these inquiries, as the Irish Data Protection Commission (DPC) has also made similar information requests. OpenAI and Microsoft are investigating whether DeepSeek used OpenAI's application programming interface (API) without authorization to train its models through a process known as distillation. An OpenAI spokesperson stated, "We know that groups in [China] are actively working to use methods, including what's known as distillation, to try to replicate advanced US AI models."

DeepSeek Suffers Its First Big Data Breach, But Your Information Is Probably Safe

Quick Links DeepSeek Leaves an Unsecured Database Online For Anyone to Find DeepSeek's Security and Privacy Issues Are a Risk For Its Users Upstart AI chatbot DeepSeek's meteoric rise has been a double-edged sword. New York-based security firm Wiz Research discovered one of the China-based company's databases exposed on the internet, containing heaps of private and sensitive data -- but your data is likely just fine. DeepSeek Leaves an Unsecured Database Online For Anyone to Find On Wednesday, 29 January 2025, Wiz Research revealed it had stumbled upon a publicly accessible database belonging to DeepSeek, the Chinese-developed AI chatbot taking the world by storm. Analysis of the database revealed a trove of DeepSeek's internal data, including user chat history and backend data, as well as sensitive data such as "log streams, API Secrets, and operational details." In addition, on accessing the DeepSeek database, the Wiz Research team gained full database control and was able to manipulate data, in turn allowing for a potential privilege escalation with DeepSeek itself. The rapid adoption of AI services without corresponding security is inherently risky. This exposure underscores the fact that the immediate security risks for AI applications stem from the infrastructure and tools supporting them. While much of the attention around AI security is focused on futuristic threats, the real dangers often come from basic risks -- like accidental external exposure of databases. As per Reuters, DeepSeek fixed the exposed database rapidly. DeepSeek's Security and Privacy Issues Are a Risk For Its Users The level of access to private DeepSeek data is a critical risk for its users. Despite DeepSeek's stratospheric rise, there are numerous questions posed around its privacy and censorship issues and how these affect its AI output. Still, some people continue to use DeepSeek despite the privacy risks, and it's not like ChatGPT, Claude, Gemini, and the other AI chatbots are privacy-protecting saints. It also highlights the issues facing DeepSeek. After surging into the world's gaze, the AI chatbot has been hit with several setbacks, including a wide-scale cyber-attack and temporarily suspending new accounts. In addition, security researchers at Kela Cyber found they could force DeepSeek to create dangerous malware and phishing campaigns while also exposing serious security flaws in its operations. So, while DeepSeek is a fantastic open-source AI model, it's a mixed bag for security, privacy, and its guardrails.

DeepSeek AI Database Exposed: Over 1 Million Log Lines, Secret Keys Leaked

Buzzy Chinese artificial intelligence (AI) startup DeepSeek, which has had a meteoric rise in popularity in recent days, left one of its databases exposed on the internet, which could have allowed malicious actors to gain access to sensitive data. The ClickHouse database "allows full control over database operations, including the ability to access internal data," Wiz security researcher Gal Nagli said. The exposure also includes more than a million lines of log streams containing chat history, secret keys, backend details, and other highly sensitive information, such as API Secrets and operational metadata. DeepSeek has since plugged the security hole following attempts by the cloud security firm to contact them. The database, hosted at oauth2callback.deepseek[.]com:9000 and dev.deepseek[.]com:9000, is said to have enabled unauthorized access to a wide range of information. The exposure, Wiz noted, allowed for complete database control and potential privilege escalation within the DeepSeek environment without requiring any authentication. This involved leveraging ClickHouse's HTTP interface to execute arbitrary SQL queries directly via the web browser. It's currently unclear if other malicious actors seized the opportunity to access or download the data. "The rapid adoption of AI services without corresponding security is inherently risky," Nagli said in a statement shared with The Hacker News. "While much of the attention around AI security is focused on futuristic threats, the real dangers often come from basic risks -- like the accidental external exposure of databases." "Protecting customer data must remain the top priority for security teams, and it is crucial that security teams work closely with AI engineers to safeguard data and prevent exposure." DeepSeek has become the topic du jour in AI circles for its groundbreaking open-source models that claim to rival leading AI systems like OpenAI, while also being efficient and cost-effective. Its reasoning model R1 has been hailed as "AI's Sputnik moment." The upstart's AI chatbot has raced to the top of the app store charts across Android and iOS in several markets, even as it has emerged as the target of "large-scale malicious attacks," prompting it to temporarily pause registrations. In an update posted on January 29, 2025, the company said it has identified the issue and that it's working towards implementing a fix. At the same time, the company has also been at the receiving end of scrutiny about its privacy policies, not to mention its Chinese ties becoming a matter of national security concern for the United States. Furthermore, DeepSeek's apps became unavailable in Italy shortly after the country's data protection regulator requested information about its data handling practices and where it obtained its training data. It's not known if the withdrawal of the apps was in response to questions from the watchdog. Bloomberg, The Financial Times, and The Wall Street Journal have also reported that both OpenAI and Microsoft are probing whether DeepSeek used OpenAI's application programming interface (API) without permission to train its own models on the output of OpenAI's systems, an approach referred to as distillation. "We know that groups in [China] are actively working to use methods, including what's known as distillation, to try to replicate advanced US AI models," an OpenAI spokesperson told The Guardian.

Report: DeepSeek's chat histories and internal data were publicly exposed

A cloud security firm found a publicly accessible, fully controllable database belonging to DeepSeek, the Chinese firm that has recently shaken up the AI world, "within minutes" of examining DeepSeek's security, according to a blog post by Wiz. An analytical ClickHouse database tied to DeepSeek, "completely open and unauthenticated," contained more than 1 million instances of "chat history, backend data, and sensitive information, including log streams, API secrets, and operational details," according to Wiz. An open web interface also allowed for full database control and privilege escalation, with internal API endpoints and keys available through the interface and common URL parameters. "While much of the attention around AI security is focused on futuristic threats, the real dangers often come from basic risks -- like accidental external exposure of databases," writes Gal Nagli at Wiz's blog. "As organizations rush to adopt AI tools and services from a growing number of startups and providers, it's essential to remember that by doing so, we're entrusting these companies with sensitive data. The rapid pace of adoption often leads to overlooking security, but protecting customer data must remain the top priority." Ars has contacted DeepSeek for comment and will update this post with any response. Wiz noted that it did not receive a response from DeepSeek regarding its findings, but after contacting every DeepSeek email and LinkedIn profile Wiz could find on Wednesday, the company protected the databases Wiz had previously accessed within half an hour.

Your DeepSeek Chats May Have Been Exposed Online

DeepSeek is having a moment: With the release of its impressive R1 model, the AI company overtook ChatGPT (and every other app) to become the number one free app on both the iOS App Store and Google Play Store. If you gave the app a try this week, however, be warned: Your chats may have been exposed. As reported by The Hacker News, DeepSeek left one of its online databases exposed. While the company has issued a fix, this database is a treasure trove of user information. It contains over one million lines of log streams, which includes chat history, secret keys (used to encrypt and decrypt data), backend information, and other important data. As of this article, DeepSeek says they are continuing to investigate the issue, despite implementing a fix on Jan. 29. It isn't clear if any parties gained access to DeepSeek's database while it was vulnerable, but the vulnerability allowed for "complete database control," as well as privilege escalation within DeepSeek's network without any authentication needed. DeepSeek's privacy and security policies have been a point of concern as so many users flock to its service. The platform collects a lot of user data, like email addresses, IP addresses, and chat histories, but also more concerning data points, like keystroke patterns and rhythms. Why does an AI app need to not only know what I typed, but how I typed it, too? As DeepSeek is a Chinese company, it stores all user data on servers in China. As such, the company is beholden by law to share any data the Chinese government requests. These practices are among the reasons the United States government banned TikTok. There's no evidence this has happened, but the whole situation paints a precarious picture for the popular AI startup. If you do want to try DeepSeek, or if you're already using it, it's important to keep these points in mind. Your user data may not be quite so secure with this particular company.

DeepSeek database was 'completely open,' leaving chat logs out there for all to see

If you're someone that's been in the tech space for the past few years, it's been a pretty wild ride since the introduction of artifical intelligence to the mainstream. While OpenAI really got the the ball rolling with the public release of ChatGPT, plenty of other companies soon followed, introducing its own tools and services in order to compete. Related Microsoft wants to know if DeepSeek obtained data from ChatGPT OpenAI thinks DeepSeek ripped them off Posts 3 For the most part, the competition has been playing catch up, with brands like Google and Microsoft playing second fiddle to OpenAI. But that doesn't mean that these same companies aren't making strides when it comes to developments in AI. Of course, these aren't the only players in the space, with brands from around the world still trying to make their mark. The hole has now been plugged Over the past week, China's DeepSeek AI has been seeing some heavy traction, with the company showing off its new R1 model that is completely free to use, while also being at the cutting edge. The big story has been that this new model isn't all that expensive to train or maintain, but some have raised questions about how DeepSeek is accomplishing these feats. While it's still unclear at this time, companies are racing to see what makes R1 tick, and because of that, researchers have discovered that the DeepSeek may not be all that secure. The Wired first reported about the security hole with the help of Wiz Research, who originally discovered the issue that leaves data exposed (via The Verge). A little digging by researchers found that it wasn't all that hard to access the data and even more alarming is that authentication was not necessary. Wiz Research shares that "this database contained a significant volume of chat history, backend data and sensitive information, including log streams, API Secrets, and operational details." And while it's unclear whether this loophole was exploited by others, the good news is that the issue has now been fixed. Of course, for these reasons and more, it's always good to be cautious about what you type into these types of products, because you don't know where the data will go and how it will be used. Some apps and services even warn users not to enter any sensitive data into prompts. With that said, this probably isn't the last time that we're going to hear about DeepSeek. Its introduction could be the start of something new, and we could even see some big changes in the space, especially when it comes to competitors like OpenAI, Nvidia, Microsoft, and others. It's definetely worth keeping an eye out for because things could really start to get interesting.

DeepSeek exposes database with over 1 million chat records

DeepSeek, the Chinese AI startup known for its DeepSeek-R1 LLM model, has publicly exposed two databases containing sensitive user and operational information. The unsecured ClickHouse instances reportedly held over a million log entries containing user chat history in plaintext form, API keys, backend details, and operational metadata. Wiz Research discovered this exposure during a security assessment of DeepSeek's external infrastructure. The security firm found two publicly accessible database instances at oauth2callback.deepseek.com:9000 and dev.deepseek.com:9000 that allowed arbitrary SQL queries via a web interface without requiring authentication. The databases contained a 'log_stream' table that stored sensitive internal logs dating from January 6, 2025, containing: "This level of access posed a critical risk to DeepSeek's own security and for its end-users," comments Wiz. "Not only an attacker could retrieve sensitive logs and actual plaintext chat messages, but they could also potentially exfiltrate plaintext passwords and local files along propriety information directly from the server using queries like: SELECT * FROM file('filename') depending on their ClickHouse configuration." Wiz says it could execute more intrusive queries but limited its exploration to enumeration to keep its research within certain ethical constraints. It is unknown if Wiz's researchers were the first to discover this exposure or if malicious actors have already taken advantage of the misconfiguration. In any case, Wiz informed DeepSeek of the matter, and the company promptly addressed the exposure, so the databases are no longer public. Apart from all the concerns that arise from DeepSeek being a China-based technology company, meaning it has to comply with aggressive data access requests from the country's government, the company does not appear to have established a solid security stance, placing sensitive data at risk. The exposure of user prompts is a privacy breach that should be very concerning for organizations using the AI model for sensitive business operations. Additionally, the exposure of backend details and API keys could give attackers a way into DeepSeek's internal networks, privilege escalation, and potentially larger-scale breaches. Earlier this week, the Chinese platform was targeted by persistent cyberattacks, which it appeared unable to thwart, forcing it to suspend new user registrations for nearly 24 hours.

DeepSeek AI exposed databases with user chat history, API keys

DeepSeek, the Chinese AI startup known for its DeepSeek-R1 LLM model, has publicly exposed two databases containing sensitive user and operational information. The unsecured ClickHouse instances reportedly held over a million log entries containing user chat history in plaintext form, API keys, backend details, and operational metadata. Wiz Research discovered this exposure during a security assessment of DeepSeek's external infrastructure. The security firm found two publicly accessible database instances at oauth2callback.deepseek.com:9000 and dev.deepseek.com:9000 that allowed arbitrary SQL queries via a web interface without requiring authentication. The databases contained a 'log_stream' table that stored sensitive internal logs dating from January 6, 2025, containing: "This level of access posed a critical risk to DeepSeek's own security and for its end-users," comments Wiz. "Not only an attacker could retrieve sensitive logs and actual plaintext chat messages, but they could also potentially exfiltrate plaintext passwords and local files along propriety information directly from the server using queries like: SELECT * FROM file('filename') depending on their ClickHouse configuration." Wiz says it could execute more intrusive queries but limited its exploration to enumeration to keep its research within certain ethical constraints. It is unknown if Wiz's researchers were the first to discover this exposure or if malicious actors have already taken advantage of the misconfiguration. In any case, Wiz informed DeepSeek of the matter, and the company promptly addressed the exposure, so the databases are no longer public. Apart from all the concerns that arise from DeepSeek being a China-based technology company, meaning it has to comply with aggressive data access requests from the country's government, the company does not appear to have established a solid security stance, placing sensitive data at risk. The exposure of user prompts is a privacy breach that should be very concerning for organizations using the AI model for sensitive business operations. Additionally, the exposure of backend details and API keys could give attackers a way into DeepSeek's internal networks, privilege escalation, and potentially larger-scale breaches. Earlier this week, the Chinese platform was targeted by persistent cyberattacks, which it appeared unable to thwart, forcing it to suspend new user registrations for nearly 24 hours.

Is Chinese AI Sensation DeepSeek a Security Risk? - Decrypt

Will China's DeepSeek AI, which became an overnight sensation, face the same kind of security scrutiny as TikTok? The company, owned by the hedge fund High-Flyer and headquartered in Hangzhou, China, is already drawing criticism for concerns about transparency and potential influence by the People's Republic of China. Social media users have been criticizing DeepSeek's AI model for refusing to answer political questions about the Chinese government and President Xi Jinping. When asked whether Xi resembles Winnie the Pooh, or what famous picture shows a man with grocery bags standing in front of tanks in Tiananmen Square, the chatbot connected to the public model, as well as the app, answers: "Sorry, that's beyond my current scope. Let's talk about something else." High-Flyer was founded in 2019 by Liang Wenfeng, an AI researcher who had initially used the nascent technology to analyze equities markets. Interestingly, his master's dissertation focused on using AI to enhance video surveillance. "Research on Target Tracking Algorithm Based on Low-Cost PTZ Camera" highlighted algorithms developed for tracking moving targets using pan-tilt-zoom cameras -- that is, cameras capable of adjusting their field of view through mechanical movements, allowing them to monitor dynamic scenes effectively. In intelligent video surveillance, automatic target tracking algorithms based on PTZ systems are crucial. These algorithms enhance traditional surveillance methods by enabling automatic detection and continuous tracking of moving objects within a scene. Users on social media were concerned about potential risks. Whether DeepSeek is surveilling its users in any shape or form is unknown. When asked whether users' queries and data are kept private, the model replies that the company "is committed to protecting user data security and privacy. We do not engage in any unauthorized form of surveillance." Still, security experts told Decrypt that the jury is still out on that question. "DeepSeek's privacy policy is unclear, and the controls in its web application aren't well known," J. Stephen Kowski, Field CTO at cybersecurity firm SlashNext Email Security+, told Decrypt. "What do they do with the data, how is it handled, where does it go, and how long is it kept? These are critical questions that need to be addressed." Indeed, Kowski attributed some of DeepSeek's rapid growth to a lack of the intense scrutiny faced by American competitors like OpenAI's ChatGPT, Google Gemini, and Anthropic's Claude AI. Infrastructure vulnerabilities have further heightened concerns about DeepSeek. Kowski highlighted potential weaknesses in the platform's code. "Validated vulnerabilities already exist, such as cross-site scripting (XSS) and prompt injection attacks that can hijack user sessions during web sessions," he said. "From what I've read, their code can potentially be manipulated to execute unauthorized commands." It should be noted, however, that users are able to download a version of DeepSeek to their computer and run it locally, without connecting to the internet. And no reports have emerged indicating that the code contains anything malicious. Meanwhile, on Monday, DeepSeek acknowledged its own security problem: It was hit with a massive cyberattack that locked new users out of the platform. "Due to large-scale malicious attacks on DeepSeek's services, we are temporarily limiting registrations to ensure continued service," the DeepSeek status page said. "Existing users can log in as usual. Thanks for your understanding and support."

Security researchers found a big hole in DeepSeek's security

They were able to access a boatload of user data in an open database. The generative intelligence platform DeepSeek , but with great popularity comes increased scrutiny. Analysts with Wiz Research have found a in the software's security. The research shows that DeepSeek left one of its critical databases exposed. This means that whoever came across the database would be allowed access to more than one million records, including user data, system logs, API keys and even prompt submissions. The researchers also noted that they were able to find the database almost immediately, without too much scanning or probing. "Usually when we find this kind of exposure, it's in some neglected service that takes us hours to find -- hours of scanning," Nir Ohfeld, the head of vulnerability research at Wiz, . But this time, he said, "here it was at the front door." Wiz Research says it's possible that a nefarious actor could have used this security hole to access other DeepSeek systems, but the company admits it only performed the base minimum assessment. This was to confirm its findings without further compromising user privacy. There is also no evidence that anyone else found the database. Wiz staffers didn't exactly know how to disclose their findings, given that DeepSeek is both a new entity and based in China. Researchers eventually sent their findings to every email address and LinkedIn profile they could find. The database was locked down within 30 minutes of the mass email. DeepSeek isn't the only AI company that has experienced a serious security breach (or two.) A hacker was able to access back in 2023 and a later that year. "AI is the new frontier in everything related to technology and cybersecurity," Ohfeld said. "Still we see the same old vulnerabilities like databases left open on the internet." As previously mentioned, DeepSeek took the world by storm in the past week or so. The disruptive AI model was allegedly created for just several million dollars. OpenAI runs through . This massive financial discrepancy sent the stock market into a tailspin, with many .

Sensitive DeepSeek database exposed to the public, cybersecurity firm Wiz reveals - SiliconANGLE

Sensitive DeepSeek database exposed to the public, cybersecurity firm Wiz reveals New York-based cloud cybersecurity firm Wiz Inc. said Wednesday it discovered a sensitive database belonging to popular Chinese artificial intelligence startup DeekSeek the company did not properly secure and thus exposed it to the public internet. DeepSeek has gained popular media notoriety in recent weeks after releasing several groundbreaking AI models, including DeepSeek-R1. According to the company, R1 rivals some of the most powerful AI models built by developers in the United States, including OpenAI's GPT-o1, and can be trained for a fraction of the cost. Wiz said that after DeepSeek gained this attention it set its sights on the company's security and "within minutes," found a publicly accessible database "completely open and unauthenticated exposing sensitive data." The database contained millions of lines of chat history, queries from users, backend data, application programming interface secrets and other sensitive operational details. "More critically, the exposure allowed for full database control and potential privilege escalation within the DeepSeek environment, without any authentication or defense mechanism to the outside world," said Gal Nagli, cloud security researcher at Wiz, in a blog post about the discovery. Nagli added that this was a major risk for the company, the level of access would have allowed attackers almost cart blanche to important information about internal systems and end-users. "Not only an attacker could retrieve sensitive logs and actual plain-text chat messages, but they could also potentially exfiltrate plaintext passwords and local files along propriety information directly from the server," he said. According to Wiz, DeepSeek quickly fixed the problem after the startup was informed about the issue. This discovery comes at a time when the popularity of DeepSeek's AI models caused a stir among investors and AI developers due to their claimed capabilities. The release of R1 caused a broad selloff in AI stocks sending Nvidia Corp.'s shares falling 17% on Monday. The sudden popularity of DeepSeek and its publicly available chatbot also made it a target of attackers who knocked registrations offline temporarily the same day. Although the attack came at the time and could have been taking advantage of the increase in registrations to tip already overburdened systems over the edge, these types of cyberattacks can also be a prelude or cover for attacks probing for cracks in infrastructure defenses. DeepSeek has also come under fire from privacy advocates and regulators for how the company collects and uses personal data. Its chatbot app has subsequently been removed from Apple Inc.'s App Store in Italy and may likely see the same fate in other countries. Italy's privacy regulator questioned if the app followed regional regulations. Fears of data leakage to the Chinese government have also compelled "hundreds" of companies, particularly those associated with governments, to block access to DeepSeek, Bloomberg reported today. Amid those concerns included weak privacy safeguards as the startup stores most of its data in China.

DeepSeek database left user data, chat histories exposed for anyone to see

Emma Roth is a news writer who covers the streaming wars, consumer tech, crypto, social media, and much more. Previously, she was a writer and editor at MUO. DeepSeek has secured a "completely open" database that exposed user chat histories, API authentication keys, system logs, and other sensitive information, according to cloud security firm Wiz. The security researchers said they found the Chinese AI startup's publicly accessible database in "minutes," with no authentication required. The exposed information was housed within an open-source data management system called ClickHouse and consisted of more than 1 million log lines. As noted by Wiz, the exposure "allowed for full database control and potential privilege escalation within the DeepSeek environment," which could've given bad actors access to the startup's internal systems. These findings were first reported by Wired. DeepSeek "promptly secured" the database after Wiz notified the startup about the issue. It's still not clear whether anyone else was able to access the exposed data, but the researchers told Wired, "it wouldn't be surprising, given how simple it was to discover." Wiz's researchers also told the outlet that DeepSeek's systems are designed similarly to those used by OpenAI, "down to details like the format of the API keys." OpenAI accused DeepSeek of using its data to train its AI models earlier this week.

Over a million lines of DeepSeek chat history was exposed in just a few minutes

Cybersecurity researchers from Wiz have found a ClickHouse database owned by Chinese AI start-up DeepSeek containing over a million lines of chat history and sensitive information. The database was publicly accessible and allowed the researchers full control over database operations. The exposure was quickly secured after Wiz shared its discovery with DeepSeek, but it's possible that information could have already been exposed. Research of this kind doesn't pry too far into the databases it finds for ethical reasons, but Wiz concluded that an attacker could potentially escalate their privileges within the DeepSeek environment and retrieve sensitive logs, chat messages, passwords, and local files -- all without needing any kind of authentication. Wiz targeted the start-up due to the recent media buzz around its R1 reasoning model, with the goal of assessing its external security. Somewhat shockingly, the ClickHouse database turned up after just a few minutes of basic searches and Wiz was able to interact with it through ClickHouses's HTTP interface. Recommended Videos From there, all the researchers had to do was run a SHOW TABLES; query, and a list of accessible datasets appeared, including the "log_steam" table that included the many lines of sensitive information. In its report, Wiz warns about the speed of AI adoption and how this pressure to develop, release, and integrate AI products as quickly as possible can lead to dangerous security practices. With all of the important and sensitive data that AI programs are now handling, the industry needs to enforce robust security practices that match those of public cloud providers and major infrastructure providers.

If You Think Anyone in the AI Industry Has Any Idea What They're Doing, It Appears That DeepSeek Just Accidentally Leaked Its Users' Chats

"The service is not mature [enough] to be used with any sensitive data at all." DeepSeek has already changed the AI game in the days since announcing its latest powerful and cheaply-trained open-source model -- but that doesn't mean the developers at the Chinese startup are infallible. Researchers at the cloud security company Wiz were poking around the back end of the groundbreaking open-source model's databases when they discovered, "within minutes," that they were able to access a trove of completely unencrypted internal data with ease. "This database contained a significant volume of chat history, backend data and sensitive information," Wiz explained in its vulnerability report, "including log streams, API Secrets, and operational details." Even worse, that wide-open back door at the open-source AI company could easily have led to an attack on DeepSeek's systems "without any authentication or defense mechanism to the outside world," the researchers wrote. As Wiz noted in its report on that glaring vulnerability, DeepSeek immediately took action to secure its databases once the security researchers alerted the company to the exposure. In conversations with Wired, however, the cloud security firm admitted that it was difficult to get in touch with anyone at DeepSeek, leaving its employees little recourse but to send LinkedIn messages and and emails to every DeepSeek-related account they could find or guess. Nobody at DeepSeek replied to Wiz's attempts at contact, but within an hour the database was locked down, Wired reports. In other words, you don't have to be competent to shake the world. The security issue doesn't sound particularly obscure, either. "Usually when we find this kind of exposure, it's in some neglected service that takes us hours to find -- hours of scanning," Wiz vulnerability research head Nir Ohfeld told Wired. With Deepseek, he said, those glaring security issues were right "at the front door." When accessing DeepSeek's databases, Wiz researchers found out a lot about the way the company's models operate, including that its infrastructure mimics OpenAI's almost exactly. This level of intelligence about a company that's exploded in popularity is relatively safe in the hands of white-hat hackers. But the researchers told Wired that if anyone else got into the DeepSeek databases before they did -- an incredibly unchallenging feat, as they indicated repeatedly -- they could have made away with as much internal data as they wanted with minimal effort. "The fact that mistakes happen is correct, but this is a dramatic mistake, because the effort level is very low and the access level that we got is very high," Wiz chief technology officer Ami Luttwak told the magazine. "I would say that it means that the service is not mature [enough] to be used with any sensitive data at all."

DeepSeek AI raises national security concerns, U.S. officials say

As Chinese AI application DeepSeek attracts hordes of American users, Trump administration officials, lawmakers and cybersecurity experts are expressing concern that the technology could pose a threat to U.S. national security. DeepSeek's introduction in the U.S. on Monday saw it quickly become the most downloaded free application in the country on Apple's app store. The rollout also rocked Wall Street as investors struggled to compute the sudden appearance of a low-cost, open-source generative AI tool able to compete with leading artificial intelligence apps such as OpenAI's ChatGPT. Shares of Nvidia, the U.S. manufacturer of advanced chips engineered for AI development, plummeted 17%, chopping roughly $600 billion off its market value -- a record single-day drop for a U.S. stock. That explosive debut was branded a "wake-up call" by President Trump on Monday. Addressing reporters on Tuesday, White House Press Secretary Karoline Leavitt said the National Security Council would examine the potential national security implications around DeepSeek's launch, noting that the administration would seek to "ensure American AI dominance." Some lawmakers also weighed in with concerns about the application's access to U.S. users. "The U.S. cannot allow Chinese Communist Party models such as DeepSeek to risk our national security and leverage our technology to advance their AI ambitions," Rep. John Moolenaar, a Missouri Republican who chairs the bipartisan House Select Committee on the Chinese Communist Party, said Tuesday in a statement shared on social media. "We must work to swiftly place stronger export controls on technologies critical to DeepSeek's AI infrastructure." The spotlight on DeepSeek comes amid rising tensions over trade, geopolitics and other issues between the two superpowers. The U.S. has already imposed significant export controls on China in an effort to rein in Beijing's production of semiconductors used in developing advanced AI, with the most recent curbs coming in December. While the calls from Moolenaar could be the first inkling of a possible congressional crackdown, Ross Burley -- a co-founder of the nonprofit Centre for Information Resilience -- warned that DeepSeek's emergence in the U.S. raises data security and privacy issues for users. Chinese law grants Beijing broad authority to access data from companies based in China. "More and more people will use it, and that will open the door to more and more personal data just being given away to the [Chinese Communist Party] and being sent basically to mainland China to be able to inform them of their activities," Burley told CBS News. "What they'll use it for is behavior change campaigns, disinformation campaigns, for really targeted messaging as to what Western audiences like, what they do," he added. DeepSeek, which is based in the Chinese city of Hangzhou, notes in its privacy policy that the personal information it collects from users is held "on secure servers located in the People's Republic of China." Under that policy, the company says it collects information including users' "device model, operating system, keystroke patterns or rhythms, IP address, and system language." DeepSeek also collects "service-related, diagnostic, and performance information, including crash reports and performance logs," according to the company. The fact that DeepSeek's servers are based in mainland China differentiates it from TikTok, the Chinese-owned social media platform that Congress had sought to ban on national security grounds before President Trump signed an executive order last week directing the Justice Department to not enforce the law for a period of 75 days. In an effort to mitigate U.S. regulatory concerns, TikTok's parent company, ByteDance, in 2022 moved all of its U.S. data to infrastructure owned by American software maker Oracle. The legislation banning TikTok -- the "Protecting Americans from Foreign Adversary Controlled Applications Act," which President Biden signed into law last April -- grants the federal government broad scope to crack down on tech platforms owned by countries regarded as U.S. adversaries. Under that law, Congress can compel a platform to divest its U.S. operations from foreign ownership, and it can be shut down if it qualifies as a threat. The law can apply to any platform that allows users to share content; has more than 1 million monthly active users; is owned by a company located in a foreign adversary-controlled country; and has been determined by the president to present a significant national security threat. But DeepSeek may be seen as less of a threat given that, unlike TikTok, it is an open-source large language model, according to Matt Sheehan, a China fellow at the Carnegie Endowment for International Peace. "[A] lot of these open-source apps, open-source models, you can actually sort of use them directly on other platforms. Perplexity is a major U.S. AI company, and they're currently using a version of DeepSeek that you can use that doesn't have the data privacy or security threats," he told CBS News. One issue that DeepSeek users face outside of China: censorship. For example, a CBS News analysis of the application found that DeepSeek did not return any results for a prompt seeking information bout the 1989 Tiananmen Square protests and subsequent massacre in Beijing. Burley of the Centre for Information Resilience thinks such suppression of information on an app being downloaded by millions of users will pressure policymakers to act. "I think it's incumbent on Western governments -- the U.K., Canada, the U.S. -- to look and see if it is wise for the Apple store and the Android store to host this large language model when it is so clearly being curated to push Chinese narratives and censorship," he said.

DeepSeek's AI avoids answering 85% of prompts on 'sensitive topics' related to China | TechCrunch

AI chatbot DeepSeek has taken the world by storm, topping app store charts and sending Silicon Valley and Wall Street into meltdown. An offshoot of Chinese hedge fund High-Flyer, DeepSeek's cost-to-performance ratio makes for impressive reading compared to incumbents such as OpenAI. However, reports have increasingly documented some of the things the AI chatbot is unwilling to discuss, including thorny topics like the Tiananmen Square protests and massacre, and Taiwan. A new report from the folks at PromptFoo, an Andreessen Horowitz-backed startup that helps companies find vulnerabilities in AI applications, has found that DeepSeek's R1 reasoning model refused to answer some 85% of 1,360 sensitive-topic "prompts," and the model's canned responses were replete with an "over-the-top nationalistic tone" often offered instead. The researchers also noted that DeepSeek can be easily jailbroken, suggesting that the Chinese lab has "implemented CCP [Chinese Communist Party] censorship in a crude, blunt-force way."

Q&A: What is China's DeepSeek and why is it freaking out the AI world?

DeepSeek, a Chinese AI startup that's just over a year old, has stirred awe and consternation in Silicon Valley after demonstrating breakthrough artificial-intelligence models that offer comparable performance to the world's best chatbots at seemingly a fraction of the cost. DeepSeek's emergence may offer a counterpoint to the widespread belief that the future of AI will require ever-increasing amounts of power and energy to develop. Global technology stocks tumbled in late January as hype around DeepSeek's innovation snowballed and investors began to digest the implications for its U.S.-based rivals and their hardware suppliers. What exactly is DeepSeek? DeepSeek was founded in 2023 by Liang Wenfeng, the chief of AI-driven quant hedge fund High-Flyer. The company develops AI models that are open-source, meaning the developer community at large can inspect and improve the software. Its mobile app surged to the top of the iPhone download charts in the U.S. after its release in early January. The app distinguishes itself from other chatbots like OpenAI's ChatGPT by articulating its reasoning before delivering a response to a prompt. The company claims its R1 release offers performance on par with OpenAI's latest and has granted licenses to individuals interested in developing chatbots using the technology to build on it. How does DeepSeek R1 compare to OpenAI or Meta AI? Though not fully detailed by the company, the cost of training and developing DeepSeek's models appears to be only a fraction of what's required for OpenAI or Meta Platforms Inc.'s best products. The much better efficiency of the model puts into question the need for vast expenditures of capital to acquire the latest and most powerful AI accelerators from the likes of Nvidia Corp. That also amplifies attention on U.S. export curbs of such advanced semiconductors to China -- which were intended to prevent a breakthrough of the sort that DeepSeek appears to represent. DeepSeek says R1 is near or better than rival models in several leading benchmarks, such as AIME 2024 for mathematical tasks, MMLU for general knowledge and AlpacaEval 2.0 for question-and-answer performance. It also ranks among the top performers on a UC Berkeley-affiliated leaderboard called Chatbot Arena. What's raising alarm in the U.S.? Washington has banned the export of high-end technologies like GPU semiconductors to China, in a bid to stall the country's advances in AI, the key frontier in the U.S.-China contest for tech supremacy. But DeepSeek's progress suggests Chinese AI engineers have worked their way around the restrictions, focusing on greater efficiency with limited resources. While it remains unclear how much advanced AI-training hardware DeepSeek has been able to access, the company's demonstrated enough to suggest the trade restrictions have not been entirely effective in stymieing China's progress. When did DeepSeek spark global interest? The AI developer has been closely watched since the release of its earliest model in 2023. Then in November, it gave the world a glimpse of its DeepSeek R1 reasoning model, designed to mimic human thinking. That model underpins its mobile chatbot app, which together with the web interface in January rocketed to global renown as a much cheaper OpenAI alternative, with investor Marc Andreessen calling it "AI's Sputnik moment." The DeepSeek mobile app was downloaded 1.6 million times by Jan. 25 and ranked No. 1 in iPhone app stores in Australia, Canada, China, Singapore, the U.S. and the UK, according to data from market tracker App Figures. Who is DeepSeek's founder? Liang, DeepSeek's founder, received bachelor's and masters' degrees in electronic and information engineering from Zhejiang University. He founded DeepSeek with 10 million yuan ($1.4 million) in registered capital, according to company database Tianyancha. The bottleneck for further advances is not more fundraising, Liang said in an interview with Chinese outlet 36kr, but U.S. restrictions on access to the best chips. Most of his top researchers were fresh graduates from top Chinese universities, he said, stressing the need for China to develop its own domestic ecosystem akin to the one built around Nvidia and its AI chips. "More investment does not necessarily lead to more innovation. Otherwise, large companies would take over all innovation," Liang said. Where does DeepSeek stand in China's AI landscape? China's technology leaders, from Alibaba Group Holding Ltd. and Baidu Inc. to Tencent Holdings Ltd., have poured significant money and resources into the race to acquire hardware and customers for their AI ventures. Alongside Kai-Fu Lee's 01.AI startup, DeepSeek stands out for its open-source approach -- designed to recruit the largest number of users quickly before developing monetization strategies atop that large audience. Because DeepSeek's models are more affordable, it's already played a role in helping drive down costs for AI developers in China, where the bigger players have engaged in a price war that's seen successive waves of price cuts over the past year and a half. What are the implications for the global AI marketplace? DeepSeek's success may push OpenAI and other U.S. providers to lower their pricing to maintain their established lead. It also calls into question the vast spending by companies like Meta and Microsoft Corp. -- each of which has committed to capital expenditures of $65 billion or more this year, largely on AI infrastructure -- if more efficient models can compete with a much smaller outlay. That roiled global stock markets as investors sold off companies like Nvidia Corp. and ASML Holding NV that have benefited from booming demand for AI services. Shares in Chinese names linked to DeepSeek, such as Iflytek Co., climbed. Already, developers around the world are experimenting with DeepSeek's software and looking to build tools with it. That could quicken the adoption of advanced AI reasoning models -- while also potentially touching off additional concern about the need for guardrails around their use. DeepSeek's advances may hasten regulation to control how AI is developed. What are DeepSeek's shortcomings? Like all other Chinese AI models, DeepSeek self-censors on topics deemed sensitive in China. It deflects queries about the 1989 Tiananmen Square protests or geopolitically fraught questions such as the possibility of China invading Taiwan. In tests, the DeepSeek bot is capable of giving detailed responses about political figures like Indian Prime Minister Narendra Modi, but declines to do so about Chinese President Xi Jinping. DeepSeek's cloud infrastructure is likely to be tested by its sudden popularity. The company briefly experienced a major outage on Jan. 27 and will have to manage even more traffic as new and returning users pour more queries into its chatbot. 2025 Bloomberg L.P. Distributed by Tribune Content Agency, LLC.

DeepSeek Is the Hot AI App. Don't Get Too Attached to It

Katie a UK-based news reporter and features writer. Officially, she is CNET's European correspondent, covering tech policy and Big Tech in the EU and UK. Unofficially, she serves as CNET's Taylor Swift correspondent. You can also find her writing about tech for good, ethics and human rights, the climate crisis, robots, travel and digital culture. She was once described a "living synth" by London's Evening Standard for having a microchip injected into her hand. AI assistant DeepSeek has rocketed to the top of the iPhone App Store free apps chart - but if you're one of the many who's downloaded the Chinese app, maybe don't get too attached. The chatbot has been lauded for its speed, efficiency and powerful reasoning skills, drawing comparisons to OpenAI's ChatGPT, which DeepSeek knocked off the top of the iPhone chart. Not only that, but it runs on less advanced chips than the top US-based competitors, raising questions about whether Silicon Valley is the leading force in AI it portrays itself as. When it comes to China-based apps and services, the first question that often arises is around user privacy. The viral success of the DeepSeek app indicates that regardless of these concerns, the AI assistant is already on many people's phones across the country. Whether you choose to trust or not trust DeepSeek with your data is totally your choice for now - but it might not be in the long run. That's because the US government is wary of Chinese technology, and could step in to interfere with the app's success in the country. There's precedent for this. You only need to look at what's currently happening with TikTok, and what previously happened to Chinese smartphone manufacturer Huawei - you can't buy a Huawei phone in the US (to say nothing of Huawei's networking gear) - to see that if Washington seriously objects to a product or service with links to China due to national security concerns, it can and will take action. Back in 2018, Huawei became the second biggest smartphone maker in the world behind Samsung, pushing Apple down into third place. But following restrictions that resulted in Huawei being forced to retreat from selling phones in the US, the company's global market share has continued to decline. Meanwhile, popular social media app TikTok has reached a crunch point in its own battle to stay active in the US, and it currently can't be downloaded on any iPhones or Android phones in the country. Unlike some chatbot rivals, the fact that DeepSeek is open source provides it with some level of protection. This means that anyone can run it on their computer and developers can tap into the API in a way that would be hard to restrict. But the DeepSeek app is still at risk. Along with potential national security concerns, the US government may well view DeepSeek as a threat to the success of homegrown companies, in the same way Huawei was to Apple and TikTok is to US-based social media companies such as Meta. DeepSeek's sudden virality coincides with what many consider to be an AI arms race, which the US hopes to dominate. Just last week, President Trump announced US-based companies OpenAI and Nvidia as key partners in his new AI infrastructure project '"Stargate" - and both companies have seen an immediate knock-on impact of DeepSeek's success on their own businesses. OpenAI's tool ChatGPT has slipped down the app store charts, while Nvidia's stock has taken a tumble. Trump hasn't publicly reacted to DeepSeek's sudden rise to prominence, but he's historically been bullish about protecting American business interests. The line between what companies supposedly constitute a national security threat and those that pose a competitive threat to US businesses can often look blurred to observers. As such, Chinese companies making waves in the US need to be squeaky clean if they want to avoid becoming the target of government scrutiny. Already, questions are being raised around DeepSeek given there seems to be censorship at play when it comes to answering questions about historical Chinese events that the country's Communist government has long sought to suppress. There's a strong chance that a blight such as this on the company's reputation could be used as a reason to restrict access in the US. Fighting the US government on a ban isn't for the weak. Both TikTok and Huawei spent significant time and resources on lobbying in the US in attempts to disprove any reason for concern. DeepSeek will need deep pockets if it ends up in a similar situation. And whether you enjoy using it and find it to be superior to ChatGPT or any other AI assistants won't matter. Public support for TikTok in the US is strong, regardless of how the country's political entities view the platform. At CNET, we've been consistently impressed by Huawei's phones over the years, finding the company's devices to be competitive with our favorite iPhones and Samsung Galaxy products. But national security concerns and political decision making about Chinese companies will always hold more sway than the quality of their products. All this is to say, if you do play around with DeepSeek, try not to get hooked. History shows us that Chinese technology can come and go from our lives within the blink of an eye, and we shouldn't rely on DeepSeek becoming another mainstay competitor in the AI assistants market - in the US at least.

China's DeepSeek is AI's hot new thing But what about censorship?

This story incorporates reporting from Business Insider, Forbes, TechRepublic and The Daily Telegraph. DeepSeek, a Chinese artificial intelligence startup, is getting significant attention in the global AI landscape with the release of its R1 model. This AI-powered chatbot has quickly positioned itself as a contender against Western counterparts like ChatGPT, Google Bard, and Meta's offerings. DeepSeek's R1 model claims to deliver advanced capabilities at a fraction of the cost of its U.S. rivals, making it an appealing option for cost-conscious users. However, the chatbot has also become a focal point for discussions on censorship due to the restrictions imposed by Chinese regulations. The release of DeepSeek's R1 model earlier this month marked a notable advancement in artificial intelligence technology. The chatbot's ascent has even caused fluctuations in the stock prices of major tech companies, indicating the potential market disruption DeepSeek poses. Users have praised the chatbot for its impressive performance, particularly in generating nuanced and conversational responses -- features that demonstrate its advanced reasoning capabilities. Despite its technological achievements, DeepSeek's R1 model is not without its limitations. One of the primary criticisms it faces is its built-in censorship protocols. Like all AI products developed in China, DeepSeek is required to adhere to the "socialist values" of the Chinese Communist Party. This means the chatbot must comply with strict government regulations, limiting discussions on sensitive topics, particularly those related to Chinese politics and governance. Such constraints highlight the broader implications of operating within the regulatory framework of China, as AI companies must navigate the delicate balance between innovation and compliance. DeepSeek's emergence has also sparked debates about the broader geopolitical implications of AI technology. The chatbot's alignment with Chinese values raises concerns about the role of AI in the dissemination of information and the potential for technology to be used as a tool for state propaganda. This is particularly relevant given the ongoing competition among global superpowers to dominate the AI industry. As DeepSeek continues to gain traction, its presence challenges the narrative predominantly controlled by U.S. tech giants, showcasing China's intent to establish itself as a leader in AI innovation. Interestingly, DeepSeek's appeal lies not just in its functionality but also in its affordability. The company behind DeepSeek has marketed the R1 model as a cost-effective alternative to American AI counterparts, raising eyebrows over its budget-friendly development. However, skepticism remains regarding the validity of these claims, especially in the absence of American-made AI chips, which are traditionally seen as essential components of high-performance AI models. As of now, DeepSeek ranks high on various metrics, including a notable position in the App Store leaderboard, signaling user enthusiasm and acceptance. By offering an option for slower but more comprehensive answers, the R1 model endeavors to meet the demands of users seeking detailed explanations. Nevertheless, the chatbot's rise underscores the complexity of marrying technological progress with political and ethical considerations. The ascent of DeepSeek highlights the evolving nature of the global tech space, characterized by rapid developments and shifting allegiances. As Chinese AI companies like DeepSeek gain prominence, the global community is prompted to reassess the dynamics of technological power and influence. Questions about the ethics of AI and the long-standing issue of censorship remain at the forefront of these discussions, challenging stakeholders to negotiate the competing demands of progress, affordability, and freedom of expression. With DeepSeek's continued innovation and expansion, it remains to be seen how the global AI industry will adapt. As China asserts itself more prominently on the world stage, the dialogue around AI's future -- both technologically and ethically -- will undoubtedly intensify.

The Censorship Dilemma Behind DeepSeek's AGI Mission

Much has been written about the AI model that jolted the stock market. But, DeepSeek CEO Liang Wenfeng rarely speaks in public, making each of his interviews and statements highly anticipated and closely scrutinised. As per reports, he appeared for just two interviews in 2023 and 2024, in which he revealed his modus operandi to achieve artificial general intelligence (AGI). Wenfeng, born in the 1980s in the Chinese province of Guangdong, graduated from Zhejiang University with a degree in electronic information engineering. In 2015, he co-founded High-Flyer, a hedge fund which managed $10 billion by 2019. The interviews highlight that, unlike many Chinese AI firms prioritising commercialisation, DeepSeek is dedicated to fundamental AGI research. "It could be two, five, or 10 years away, but it will definitely happen in our lifetime," he said, focusing on three main directions: mathematics and code, multimodality, and natural language itself. Elaborating on DeepSeek's approach to talent, Wenfeng clarified that there are no "wizards". According to him, the company operates with a bottom-up structure, recruiting young local talent from local Chinese universities. While well-funded, DeepSeek's main hurdle lies in securing high-end chips restricted by US export controls. "We don't have short-term fundraising plans. Our problem has never been funding; it's the embargo on high-end chips," he said. According to The Wall Street Journal, Wenfeng recently met with Chinese premier Li Qiang to discuss the difficulties Chinese companies face as a result of US restrictions on advanced chip exports. On the open-source front, Wenfeng said, "In the face of disruptive technology, a closed-source moat is temporary." He also noted that while people often speak of a one or two-year gap between Chinese and American AI, the true divide is between originality and imitation. As DeepSeek rattles global markets, it also raises serious concerns about AI safety, driven by its open-source design and strong links to the Chinese Communist Party. This is especially true, considering top leaders predict that ASI is on an accelerated timeline from before. Censorship of sensitive topics is a major concern with DeepSeek. The model avoids answering questions related to issues such as Uyghur human rights abuses, Taiwan's political status, the 1989 Tiananmen Square incident, criticism of Chinese supreme leader Xi Jinping, censorship in China, and questions about Arunachal Pradesh and Kashmir's sovereignty, among others. Instead, it deflects these inquiries with responses like: "Sorry, I'm not sure how to approach this type of question yet." People on X have DeepSeek' compared the censorship on Anthropic's Claude and OpenAI's ChatGPT. According to some reports, China's regulatory body, the Cyberspace Administration of China (CAC), imposes strict testing requirements for AI models, including testing up to 70,000 questions to ensure politically safe answers. This slows AI development and limits the randomness and creativity typical of generative AI. The Chinese government rigorously reviews large language models (LLMs) to ensure they adhere to "core socialist values". Companies such as ByteDance, Alibaba, Moonshot, and 01.AI are obligated to undergo these compulsory audits conducted by the CAC. Moonshot's chatbot, Kimi, rejects most questions about Jinping. Similarly, ByteDance's LLM ranks highest in safety compliance tests, showcasing its alignment with Beijing's messaging. Many argue that because it's open-source, it can be fine-tuned to suit specific needs or values. But this also hint at a deeper issue; one of censorship that extends beyond simple fine-tuning. Former OpenAI researcher Miles Brundage pointed out that while this is an immediate benefit, it could lead to stricter rules in the future. DeepSeek and governments might focus more on improving AI safeguards, making it harder to release new models. Governments could also push for AI tracking on devices, where smaller AIs monitor the use of larger ones and send reports to central systems. This could also affect how non-Chinese AI models are used in China.

What is DeepSeek, the Chinese AI app challenging OpenAI and Silicon Valley?

The Chinese AI app's success with U.S. consumers is shaking up assumptions about the global AI race. DeepSeek is a new artificial intelligence chatbot that's sending shock waves through Wall Street, Silicon Valley and Washington. The app, name after the Chinese start-up that built it, rocketed to the top of Apple's App Store in the United States over the weekend. According to some experts, DeepSeek's success and a technical paper it published last week suggest that Chinese AI developers can match their U.S. counterparts without the same access to expensive, high-end hardware that is at the heart of geopolitical tensions between Washington and Beijing. Here's what you should know about the DeepSeek AI assistant app and the company behind it. What to know What is DeepSeek?How does DeepSeek work?Is it safe to use?How did DeepSeek do it?What makes DeepSeek different from ChatGPT?How have markets and Silicon Valley reacted to DeepSeek? Show all questions

DeepSeek is taking over the AI world -- live updates

There's a new AI game in town called DeepSeek. Owned a by the Chinese stock trading firm High-Flyer, it has risen in popularity at an alarming rate, putting other AI chatbots like Google Gemini and ChatGPT on notice. DeepSeek's AI reportedly required far less raw computing power to get started than rivals from larger tech firms. In fact, the company said it only needed about $6 million in computing power to train its new system, approximately 10 times less than what Meta spent training its AI model. The DeepSeek is currently No. 1 in the Apple App Store and shows no signs of slowing down -- beyond outages due to demand. What does this mean for you and what are the latest developments around DeepSeek? Should we be worried about it being banned, similar to what happened to Chinese-owned TikTok? We'll investigate all that and more in our live coverage on DeepSeek.

DeepSeek: Is China's AI tool as good as it seems?

DeepSeek looks and feels like any other chatbot, though it leans towards being overly chatty. Just as with OpenAI's ChatGPT or Google's Gemini, you open the app (or website) and ask it questions about anything, and it does its best to give you a response. It gives long answers and will not be drawn on expressing an opinion, however directly it is asked for one. The chatbot often begins its response by saying the topic is "highly subjective" - whether that is politics (is Donald Trump a good US president?) or soft drinks (which is more tasty, Pepsi or Coke?). It wouldn't even commit to saying whether or not it was better than OpenAI's rival artificial intelligence (AI) assistant ChatGPT, but it did weigh up the pros and cons of both - ChatGPT did exactly the same, and even used very similar language. DeepSeek says it was trained on data up to October 2023, and while the app seems to have access to current information such as today's date, the website version does not. That is not dissimilar to earlier versions of ChatGPT and is probably a similar attempt at safeguarding - to stop the chatbot spewing out misinformation pumped onto the web in real time. It can be quite fast in its responses, but is currently groaning under the weight of so many people rushing to try it out as it has gone viral. But there is one area in which it is nothing like its US rival - DeepSeek censors itself when it comes to questions about subjects banned in China.

DeepSeek Has a Very Interesting Answer About the Tiananmen Square Massacre

Chinese AI startup DeepSeek has made a huge splash with its ChatGPT competitor, claiming it developed a similarly-performing AI assistant at a fraction of the cost. It's a serious contender -- at least in the eyes of investors, with AI chipmaker Nvidia's shares sliding by around 15 percent Monday morning. But the app also has some significant shortcomings. Like other Chinese AI models, DeepSeek is beholden to the rules of state censors, as Bloomberg reports, refusing to directly address sensitive topics like the 1989 Tiananmen Square massacre or China-Taiwan relations. Users on Reddit found that when asked about a "famous picture" of a "man with grocery bags in front of tanks," the large language model starts to oblige, producing text about the world-famous "Tank Man" photograph taken on the day of the Tiananmen Square protests. But moments later, the text disappears and is replaced with a generic error message, as seen in a screen recording shared by one Reddit user, seemingly showing how the app is (however sloppily) abiding by Chinese censorship rules. "Sorry, that's beyond my current scope," the message reads. "Let's talk about something else." As another user on Reddit found, even asking it more broadly about criticisms of the Chinese government, it had an intriguing answer. "Sorry, I'm not sure how to approach this type of question yet," it wrote. "Let's chat about math, coding, and logic problems instead!" The user got the same response when they included the name of Chinese president Xi Jinping in their query. "I literally just asked it if Taiwan was its own country then it answered like [ChatGPT] would then right after replaced it with a message like 'I'm not programmed to answer that kind of question yet' or some such," one user wrote. One user on Hacker News claims to have circumvented these censorship rules by asking the chatbot for a detailed breakdown of what happened in 1989, resulting in an elaborate answer. While China's "Great Firewall" censorship rules are well-known and established, OpenAI's ChatGPT also has plenty of guardrails limiting what it can say. Those may be retreating somewhat, though; earlier this month, OpenAI was caught quietly removing language endorsing "politically unbiased" AI, a design purportedly inspired to "streamline" its documentation, as TechCrunch reported at the time. Meanwhile, conservatives have long accused AI companies of censoring their viewpoints, with multi-hyphenate billionaire Elon Musk founding an AI startup called xAI to develop an "anti-woke" AI (with less-than-stellar results). Last week, president Donald Trump signed an executive order, calling on the AI industry to "develop AI systems that are free from ideological bias or engineered social agendas." Whether that'll lead to the loosening of guardrails -- or the establishment of a new set of state censorship rules to further a far-right agenda -- for US-based chatbots like ChatGPT remains to be seen.

DeepSeek's new AI chatbot and ChatGPT answer sensitive questions about China differently

HONG KONG (AP) -- Chinese tech startup DeepSeek 's new artificial intelligence chatbot has sparked discussions about the competition between China and the U.S. in AI development, with many users flocking to test the rival of OpenAI's ChatGPT. DeepSeek's AI assistant became the No. 1 downloaded free app on Apple's iPhone store on Tuesday afternoon and its launch made Wall Street tech superstars' stocks tumble. Observers are eager to see whether the Chinese company has matched America's leading AI companies at a fraction of the cost. The chatbot's ultimate impact on the AI industry is still unclear, but it appears to censor answers on sensitive Chinese topics, a practice commonly seen on China's internet. In 2023, China issued regulations requiring companies to conduct a security review and obtain approvals before their products can be publicly launched. Here are some answers The Associated Press received from DeepSeek's new chatbot and ChatGPT: For many Chinese, the Winnie the Pooh character is a playful taunt of President Xi Jinping. Chinese censors in the past briefly banned social media searches for the bear in mainland China. ChatGPT got that idea right. It said Winnie the Pooh had become a symbol of political satire and resistance, often used to mock or criticize Xi. It explained that internet users started comparing Xi to the bear over similarities in their physical appearances. DeepSeek's chatbot said the bear is a beloved cartoon character that is adored by countless children and families in China, symbolizing joy and friendship. Then, abruptly, it said the Chinese government is "dedicated to providing a wholesome cyberspace for its citizens." It added that all online content is managed following Chinese laws and socialist core values, with the aim of protecting national security and social stability. It might be easy for many people to answer, but both AI chatbots mistakenly said Joe Biden, whose term ended last week, because their data was last updated in October 2023. But they both tried to be responsible by reminding users to verify with updated sources. The 1989 crackdown saw government troops open fire on student-led pro-democracy protesters in Beijing's Tiananmen Square, resulting in hundreds, if not thousands, of deaths. The event remains a taboo subject in mainland China. DeepSeek's chatbot answered, "Sorry, that's beyond my current scope. Let's talk about something else." But ChatGPT gave a detailed answer on what it called "one of the most significant and tragic events" in modern Chinese history. The chatbot talked about the background of the massive protests, the estimated casualties and the legacy. DeepSeek's chatbot's answer echoed China's official statements, saying the relationship between the world's two largest economies is one of the most important bilateral relationships globally. It said China is committed to developing ties with the U.S. based on mutual respect and win-win cooperation. "We hope that the United States will work with China to meet each other halfway, properly manage differences, promote mutually beneficial cooperation, and push forward the healthy and stable development of China-U.S. relations," it said. ChatGPT's answer was more nuanced. It said the state of the U.S.-China relationship is complex, characterized by a mix of economic interdependence, geopolitical rivalry and collaboration on global issues. It highlighted key topics including the two countries' tensions over the South China Sea and Taiwan, their technological competition and more. "The relationship between the U.S. and China remains tense but crucial," part of its answer said. Again -- like the Chinese official narrative -- DeepSeek's chatbot said Taiwan has been an integral part of China since ancient times. "Compatriots on both sides of the Taiwan Strait are connected by blood, jointly committed to the great rejuvenation of the Chinese nation," it said. ChatGPT said the answer depends on one's perspective, while laying out China and Taiwan's positions and the views of the international community. It said from a legal and political standpoint, China claims Taiwan is part of its territory and the island democracy operates as a "de facto independent country" with its own government, economy and military. Associated Press writer Ken Moritsugu in Beijing contributed to this story.

Donald Trump urged to 'ban' new Chinese AI app

A leading tech expert has demanded US President Donald Trump ban new Chinese artificial intelligence start-up DeepSeek that has rocked the tech world and wiped almost $1 trillion from the US stock market. DeepSeek was released last week and uses cheaper microchips than OpenAI, which runs ChatGPT and uses chips from powerhouse Nvidia. The emerging AI competitor has also raised eyebrows over its responses to questions about the Chinese President Xi Jinping, China's treatment of Uyghur Muslims, the nation's fraught relationship with Taiwan or the 1989 Tiananmen Square protests and massacre. Technology expert and commentator Trevor Long on Tuesday raised concerns about the new player and called on the newly inaugurated US President to take action. "The real story today will evolve to be the Chinese government and the Chinese ownership of this app," Mr Long told Sky News. "If we're gonna have these conversations about TikTok - if I was Donald Trump today I would ban this app." Mr Long said he also experienced bizarre responses from DeepSeek when he was inquiring about the controversial issues that had been widely censored in China. "I have ... had a couple of little plays here and said, 'What happened to Tiananmen Square in 1989?' 'Sorry, I can't discuss that'," Mr Long said. "'Is the Chinese government able to access my data?' It gave me a long response and then deleted that response and said, 'Let's talk about something else'. "It is mind blowing the censorship that exists within this app if you try and talk about Chinese issues." Concerns have also been raised about DeepSeek's claim it developed its new AI model 'R1' for just $5.6 million, which is miniscule compared to the billions poured into the US equivalents. AI models from ChatGPT to DeepSeek require advanced chips to power their training. The Biden administration had, since 2021, widened the scope of bans designed to stop these chips from being exported to China. The new players claims to have just 2050 powerful H100 chips from Nvidia, but tech billionaire Elon Musk casted doubt on this. Musk made his claim in response to a video where Scale AI CEO Alexander Wang claimed there were more highly powerful H100 chips in the Chinese labs, which China was not meant to have, than they let on. "The Chinese labs, they have more H100s than people think ... my understanding is DeepSeek has about 50,000 H100s - which they can't talk about obviously because it is against the expert controls that the United States has put in place," Mr Wang said. The X owner, who founded his social media platform's own artificial intelligence venture xAI in March 2023, backed in Mr Wang's call. "Obviously," Musk wrote. DeepSeek is currently the most downloaded app in the US and there have been outages on the website after it was reportedly hit with large-scale cyberattacks. "Due to large-scale malicious attacks on DeepSeek's services, we are temporarily limiting registrations to ensure continued service," the company's statement read. "Existing users can log in as usual." Mr Long said the emergence of this AI operator sparked concerns about what these major players were doing with its users' data. "The real controversy around AI is what information was used to train it, to educate it, and secondly, what happens with the information that it is fed," he said. "For example, I can upload up a government document (such as) a ten-page report and say, 'Can you summarise this for me?' But what's happening to that document? Where is it being stored?" DeepSeek's emergence sent Nvidia stock plummeting on Monday, dropping an astonishing 17 per cent after the opening bell and stripping $US600 billion ($955 billion) from the company's market capital.

Could Trump ban DeepSeek? What the TikTok ban saga tells us.

What's next for DeepSeek, politically? Credit: Andrey Rudakov/Bloomberg via Getty Images DeepSeek, a Chinese open-source AI system similar to ChatGPT, has risen to popularity at a peculiar time: in the midst of an ongoing legal battle over whether another Chinese tech platform, TikTok, should be allowed to run in the U.S. Some users are curious if the U.S. government would attempt to ban DeepSeek on the same grounds it has used to attempt to ban TikTok. In short, sure, the U.S. could ban DeepSeek if it wanted to. It has the capacity to ban things it doesn't like from countries it doesn't trust in order to protect its citizens' data. In the case of TikTok, lawmakers who voted in support of banning the app cited concerns about data privacy, national security, surveillance, and propaganda, primarily due to the app's Chinese ownership. These lawmakers argue that TikTok is controlled by a "foreign adversary" -- in this case, its Chinese parent company, ByteDance -- and it isn't in the U.S.'s interest to allow foreign adversaries access U.S. citizens' data. TikTok has routinely denied allegations that it shares U.S. users' data with the Chinese government. This line of thinking could lay out some problems for DeepSeek, too. The company is based out of China. According to its privacy policy, "The personal information we collect from you may be stored on a server located outside of the country where you live. We store the information we collect in secure servers located in the People's Republic of China." But the TikTok ban wasn't all purely due to privacy concerns -- there were lobbying efforts afoot, too. Beyond that, there are also similar players involved in DeepSeek's rise to power. Meta and YouTube stand to benefit from a TikTok ban, and those tech companies, along with OpenAI and others, would also probably benefit from a ban on DeepSeek, which would let them better corner the AI market. Those power players aren't making any explicitly public moves just yet. That said, President Donald Trump doesn't seem too concerned with the privacy issues at TikTok currently -- at least, not enough to put the ban into effect -- so it's unclear if he'll follow that same logic regarding DeepSeek. He did, however, sign a whole host of executive orders during the first week of his presidency, including one that tackles AI. The order revokes a Biden AI executive order "which hampered the private sector's ability to innovate in AI by imposing government control over AI development and deployment" and "calls for departments and agencies to revise or rescind all policies, directives, regulations, orders, and other actions taken under the Biden AI order that are inconsistent with enhancing America's leadership in AI." He did keep one Biden-era order that designates more federal land for AI data centers, NPR reported. This works alongside the Stargate project that OpenAI announced alongside Trump last week. As former Rep. Patrick McHenry said on CNBC on Monday, banning an open-source model like DeepSeek wouldn't necessarily "bring the globe to our regime, our rule of law, speech rights regime, open-society regime," a theory that lines up with some legislators' take on the TikTok ban. "It actually leads the rest of the world to go a different direction for us." "The president's capacity to take action here is massive against an app or an adversary he doesn't like," Former Rep. McHenry said. "We see this in tariffs; we're seeing this in tariffs moment to moment, hour to hour, and that will continue. But when it comes to export controls, the authorities of this administration, like the last one, are very broad, very deep, and quite meaningful." Right now, it's unclear what the government might do regarding DeepSeek, but it has options.