3DNews AI
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
White House makes AI incident reporting mandatory after Anthropic episode
After September incidents in which Anthropic models sent data on their own through US government websites, the White House replaced voluntary reporting with mandatory reporting.
OpenAI Fires Three Safety Researchers Over Contacts With Outside Organizations
Three former OpenAI employees say they were fired for talking to outside AI safety organizations. The company confirmed the dismissals but gave a different reason.
SemiAnalysis: Chinese AI Labs Rarely Publish Model Safety Test Results
SemiAnalysis finds that Chinese AI developers are reluctant to disclose how they test models for dangerous scenarios, even as the world debates the risks of frontier systems.
Researchers find discrepancies in OpenAI's Lean proof on Navier–Stokes
Comparing OpenAI's Navier–Stokes proof with its Lean version turned up weakened claims at several key points. It is not a refutation, at least not yet.
AI Agents Make Costlier Mistakes Than Chatbots: They Leak Data and Buy Without Asking
AI agents take over routine work, but their failures hit wallets and reputations: from posting confidential data in a work chat to making purchases without access to payment methods.
Anthropic adds two tools to Claude: dashboards and animations in chat
Anthropic has released two tools for business users in Claude: real-time data dashboards and animations that are built right in the chat window.
Anthropic Launches Free OSS Scanner to Hunt for Open Source Vulnerabilities
Anthropic has launched OSS Scanner — a free service for automatically finding vulnerabilities in open source projects. It uses Claude Mythos and other powerful models. Reports are generated entirely by AI, so developers
Google lets the Pentagon use its AI models for any lawful purpose
Google is the Pentagon's newest AI contractor: the US military can use its models in classified systems and for any purpose the law allows.
Pentagon makes Palantir's Maven the core AI system for the US Army
The US Department of Defense has chosen Palantir's Maven as its main combat AI platform, standardizing it across the army, intelligence and future military AI work.
Biological AI as Weapon and Shield: Scientists Search for Balance
Specialized AI models make it easier to design pathogens and toxins, but could in theory help build defenses. Scientists in Nature disagree on control: restrict access, or monitor DNA synthesis?
Meta Starts Tracking Employees' Mouse Movements, Sparking Mass Protests
Meta deployed an app that tracks mouse movements on work PCs, angering US staff over privacy and layoff risks. *Meta is designated an extremist organization and banned in Russia.
Students Can No Longer Read Long Texts, and Instructors Blame AI
A US instructor found his students can no longer get through a 20-page article. They lose focus and can't absorb complex texts, and the problem is now systemic.
Meta Is Stuck on Advertising: Why Diversification Isn't Working
Despite every attempt to diversify, Meta Platforms still gets 97.6% of its revenue from ads, a classic trap for big tech. *Meta is designated an extremist organization and banned in Russia.
SpaceX is leasing out all of its Colossus 1 data center after its engineers struggled to run it
SpaceX couldn't run its Colossus 1 data center in Tennessee efficiently and has decided to lease the entire facility, despite an acute shortage of compute on the market.
OpenAI spread itself across everything — and missed AI coding, which Anthropic claimed
OpenAI, which launched ChatGPT in November 2022, chased every direction at once and missed the market for AI developer tools, which Anthropic has taken.
AI Agents Outperform Human Scammers at Social Engineering, International Study Finds
International researchers show that AI agents trained in social engineering win victims' trust more effectively than human scammers, extending the threat beyond technical attacks.
Yandex B2B Tech to Release Universal AI Agent for Business in September 2026
Yandex B2B Tech, Yandex's enterprise arm, will launch a single AI agent for office tasks in September 2026, replacing today's patchwork of corporate AI tools.
Elon Musk: Tesla's AI6 Chip Will Be the World's Best for Edge AI Computing
Elon Musk says Tesla's upcoming AI6 chip will be the world's best for edge AI computing, serving as the foundation for robotaxis, Optimus robots and Autopilot.
Apple is preparing an AI shopping assistant: a section about it turned up in the Apple Store privacy policy
Apple added a section on an unreleased AI shopping assistant to the Apple Store app's privacy policy, the first official mention of the assistant.
Laid-Off Meta Employees Fail to Prove in Court That AI Was Used to Pick Layoffs
Laid-off Meta employees accuse the company of using discriminatory AI to build layoff lists, but proving the technology was actually used has turned out to be nearly impossible in court. *Meta is recognized as an extrem
AI Boom Revives Demand for Hard Drives as HDD Component Makers Expand Capacity
The AI boom has revived demand for classic hard drives: servers and data centers need cheap bulk storage, and HDD component makers are expanding capacity.
Apple and Broadcom extend custom AI chip supply deal through 2031
Apple and Broadcom have extended their deal to develop and supply custom AI chips through 2031, locking in Broadcom's production capacity for years to come.
Apple is in talks with PrismML to compress AI models like Qwen 3.6 for the iPhone
Apple is discussing PrismML's LLM compression tech with the startup to run powerful AI directly on the iPhone without the cloud, The Information reports.
Yandex's Alice AI can now describe what its smart IP cameras see
Owners of Yandex smart IP cameras can now ask the Alice AI assistant what's happening on camera via chat, a smart speaker, TV Station, or Dropes earbuds.
SpaceXAI (formerly xAI) unveils Grok 4.5, built for coding and agents
SpaceXAI, formerly xAI, has unveiled Grok 4.5, which it calls its most powerful model yet, tuned for coding and agentic developer tasks.
Investors Wiped $2.3 Trillion Off the 'Magnificent Seven' Tech Giants in June 2026
Investors doubted Big Tech's massive AI spending would pay off and cut $2.3 trillion from the Magnificent Seven's market value in June 2026.
Anthropic unveils Claude Science, an AI model built to speed up scientific research and drug discovery
Anthropic has released Claude Science, a specialized model meant to help scientists fight diseases and develop new drugs faster.
Meta says its new Watermelon model has nearly caught up with OpenAI's GPT-5.5
Meta AI chief Alexandr Wang said a new model codenamed Watermelon has nearly matched the capabilities of OpenAI's flagship GPT-5.5, Business Insider reports.
Zhipu AI's Z.ai model matches Anthropic at finding vulnerabilities in code
Zhipu AI's Z.ai model has matched Anthropic's Mythos at finding code vulnerabilities but still trails Anthropic and OpenAI on other tasks, WSJ reports.
Court dismisses Elon Musk's xAI lawsuit accusing OpenAI of stealing Grok trade secrets
A court rejected xAI's claims against OpenAI. xAI alleged OpenAI tried to access its trade secrets by poaching a Grok developer.