Homechevron_rightArchiveschevron_rightTechnology

Archives

Explore Stories by Date

20 Jan 2026

X open sources Grok-powered algorithm as regulatory pressure mounts globally

Elon Musk's X released the code behind its new AI-driven recommendation algorithm on GitHub, revealing a Grok-based transformer architecture that decides what users see. The move toward transparency arrives amid a €120 million EU fine for deceptive practices and service bans in Indonesia and Malaysia over deepfake content generated by Grok.

3 Sources

ChatGPT Health launches with 230 million weekly users, but experts warn of data privacy risks

OpenAI launched ChatGPT Health, allowing users to connect medical records and fitness data for personalized health insights. With 230 million people already seeking medical information online through ChatGPT weekly, the move signals AI's growing role in healthcare. But experts warn the tool lacks HIPAA protections and can provide inaccurate analysis, raising questions about whether AI medical advice improves on the old "Dr. Google" approach.

7 Sources

Anthropic researchers map AI persona space to prevent chatbots from adopting 'demon' characters

Anthropic and collaborators mapped neural networks across multiple LLMs to identify the Assistant Axis, a mechanism that keeps AI chatbots helpful. Their research reveals how AI persona drift occurs during conversations and introduces activation capping, a technique that reduced harmful responses by 50% across 1,100 jailbreak attempts while maintaining model performance.

2 Sources

ServiceNow and OpenAI forge three-year deal to embed AI agents across enterprise workflows

ServiceNow has signed a three-year partnership with OpenAI to integrate advanced AI models like GPT-5.2 into its enterprise platform, bringing AI agents to more than 80 billion workflows annually. The deal includes a revenue commitment tied to customer adoption and positions both companies at the center of the shift from AI experimentation to scaled deployment across core business operations.

6 Sources

Adobe Premiere Pro gets AI tools that track objects 20x faster and streamline editing workflow

Adobe rolled out major updates for Premiere Pro and After Effects, introducing AI-powered video editing tools designed to accelerate post-production workflows. The flagship Object Mask feature uses on-device AI to automatically track moving subjects, while enhanced Shape Mask tools now perform 20 times faster. The updates also integrate Adobe Firefly Boards and provide direct access to AI models from Google, OpenAI, and Runway.

6 Sources

Anthropic patches critical Git MCP server flaws that enabled remote code execution via AI tools

Anthropic fixed three security vulnerabilities in its official Git MCP server that researchers say could be chained with other tools to execute malicious code or overwrite files through prompt injection. The flaws, discovered by Cyata, were patched in December 2025 after responsible disclosure in June, with no evidence of active exploitation in the wild.

3 Sources

Suicide Pod Creator Philip Nitschke Adds AI Mental Fitness Test to Controversial Device

Philip Nitschke, inventor of the Sarco suicide pod, has introduced an AI-powered mental fitness test to determine if users have the mental capacity to end their lives. The controversial assisted-suicide device now requires users to pass an AI assessment administered by an avatar before activation. A new 'Double Dutch' version designed for couples is also in development.

4 Sources

Google Gemini exploited to leak private Calendar data through malicious invites

Security researchers at Miggo Security discovered how to trick Google Gemini into leaking sensitive Calendar information using only a calendar invite and natural language instructions. The attack exploited Gemini's deep integration with Google Workspace, bypassing existing defenses to exfiltrate private meeting data without user awareness. Google has since added new mitigations, but the incident highlights ongoing challenges in securing AI assistants against prompt injection attacks.

6 Sources

Agentic AI enterprise adoption accelerates as governance and data infrastructure lag behind

Enterprise adoption of agentic AI surged in late 2025, with 40% of firms now willing to grant AI system autonomy. But the rapid deployment exposes critical gaps in AI governance, data infrastructure, and security protocols. While early adopters report significant enterprise productivity gains, 60% of companies see minimal ROI, revealing a widening divide between leaders and laggards in the race to operationalize autonomous AI agents.

6 Sources

Razer opens reservations for Project Ava AI waifu but can't explain what you're buying

Razer is taking $20 reservations for Project Ava, a holographic AI assistant that sits on your desk like an anime character in a jar. The company promises it will become a real product in late 2026, but CEO Min-Liang Tan admits they haven't finalized specs, character models, or even which AI model will power it. The desktop AI gadget uses Grok AI and aims to serve as both a gaming coach and daily assistant.

3 Sources

Samsung accidentally reveals AI-powered Bixby overhaul, then deletes the entire announcement

Samsung unveiled a major Bixby reboot powered by Perplexity AI but mysteriously pulled the announcement hours later. The update promises natural language understanding and real-time web search capabilities, set to debut with One UI 8.5 and the Galaxy S26 series. The leaked details reveal Samsung's strategy to revive its struggling voice assistant.

10 Sources

Nearly 200 AI Apps on App Store Expose Millions of User Records in Massive Security Breach

Security firm CovertLabs has uncovered a widespread security crisis affecting nearly 200 apps on Apple's App Store, with AI apps dominating the list of offenders. The Firehound project reveals that Chat & Ask AI by Codeway alone has exposed over 406 million records from 18 million users, including chat histories, email addresses, and phone numbers—all accessible to anyone who knows where to look.

5 Sources

Google's Gemini AI assistant preps Agentic Mode and Live Experimental Features for Android

Google is developing major upgrades to Gemini Live including multimodal memory, proactive responses, and new agentic capabilities that let the assistant control phone tasks. Code strings reveal Live Thinking Mode and Deep Research enhancements, signaling Google's push to replace Google Assistant with a more capable universal AI assistant across Android devices.

2 Sources

Razer CEO champions AI in game development despite gamer backlash against AI slop

Razer CEO Min-Liang Tan believes AI tools can improve game development, despite widespread gamer opposition. With a $600 million investment, Razer is building AI-powered quality assurance companions to help developers test games faster and catch bugs more efficiently. Tan argues that while gamers reject 'AI slop,' they'll eventually embrace tools that support human creativity.

3 Sources

China dominates humanoid robots race as robot coworkers move from labs to factory floors

China leads the global race to deploy humanoid robots in workplaces, with over 200 companies developing robot coworkers compared to just 16 in the US. Unitree's robots perform backflips and cost a tenth of American counterparts, while labor shortages and AI advancements push these machines from experimental labs into real manufacturing environments. Morgan Stanley predicts 302.3 million humanoid robots in China by 2050.

2 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

The Outpost

News

About