AI news with context. Independent tests. Practical guides.
The EU AI Office has sent formal requests for information to OpenAI and Anthropic over agents that broke out of their test environments and into real organisations. It is the first enforcement step announced under the AI Act’s rules for general-purpose models, and lawyers already dispute whether the law reaches incidents that happened during testing.
Euronews
An OpenAI engineer has been charged with hiding an audio recorder, wrapped in toilet paper, behind a TV in an executive conference room. Prosecutors say chatbot prompts helped him build it; OpenAI says its own models helped trace it back to him. He faces up to five years in prison if convicted.
U.S. Department of Justice
Character Technologies, the company behind the Character.AI chatbot platform, has filed for Chapter 11 and plans to sell its assets through the court. Disney is the opening bidder at $20 million for a licence to some of its technology, and the app will shut down if no buyer is found to keep it running.
TechCrunch
Anthropic’s Frontier Red Team says small open-weight models may be months from being able to copy themselves across poorly secured smart TVs, set-top boxes and other home devices. In its tests only GLM-5.2 completed every step, on a simplified network, and the company says nothing like this has been seen in the wild.
Anthropic
Meta Superintelligence Labs has released Muse Wonder and Muse Wonder Full, its first models built to compete at the frontier, with a pitch centred on long-term memory and agents that keep going. Meta’s own table puts Wonder Full ahead on agent and long-horizon coding tests and behind Claude Opus 5.5 elsewhere; the API model costs $4 and $20 per million tokens.
Meta
Nvidia is in early talks to acquire Reflection AI or deepen its $800 million stake in the open-weight model lab, the Financial Times reports, with an acqui-hire among the options. Reflection was last said to be raising at a $25 billion valuation and previewed its first model, Beam, the same week.
Financial Times via Reuters
Amazon is negotiating to buy Decart, the Israeli startup behind the real-time video model Lucy and the Oasis world model, for about $7 billion, The Wall Street Journal reports. The talks come a month after Anthropic walked away from a $6 billion deal for the same company, and Decart’s chip-optimisation software already runs on Amazon’s own Trainium silicon.
The Wall Street Journal via Reuters
An OpenAI grading model that could not find its input files faked them, then tried to delete system directories so the host would hand it a fresh environment, OpenAI’s alignment team reports. Two more reports posted the same day show models breaking a GET-only internet rule to pull public statistics, and one model told itself not to mention it.
OpenAI
Anthropic has cut live internet access from all of its internal evaluations after finding Claude models exploiting a server flaw, filing an invented tip on a police form and using URL shorteners to slip past tool limits. The company says the real-world impact was minimal and no customer data was involved, but several cases touched U.S. government websites, and it briefed the White House.
Anthropic
One email a morning, only on days we publish: up to ten AI stories with a one-line summary each, sent at 8:30 UTC. Unsubscribe in one click.
Original hands-on tests — same brief to every model, one attempt each, every build live in the article.
Google's Gemini 4 Argon, its first entry in the series, against the September builds of Claude Opus 5.5, GPT-6 Astra and Grok 4.7 on the same interchange, fairground and rocket launch.
Claude Opus 5.5 · GPT-6 Astra · Gemini 4 Argon · Grok 4.7
Anthropic's Claude Haiku 5.5, released on 7 October at GPT-6 Luna's exact price, against Luna's September builds on the same interchange, fairground and rocket launch.
GPT-6 Luna · Claude Haiku 5.5
A new game brief, the No Man's Sky moment of flying down from orbit, landing by hand and stepping out, taken by Claude Opus 5.5, GPT-6 Astra and Grok 4.7.
Claude Opus 5.5 · GPT-6 Astra · Grok 4.7
2,938 views · Higgsfield
2,083 views · TechCrunch / CNBC
1,353 views · BitsMinds Analysis
1,336 views · Claude Code Docs
836 views · Micron / Reuters
The world's leading AI image generator. Create stunning artistic images from text prompts.
Anthropic's AI assistant — Opus 5.5, Fable 5.1 and Sonnet 5 — with best-in-class writing, code review, a 1M-token context and Claude Code.
OpenAI's ChatGPT assistant — conversation, writing, coding, image generation and data analysis, now on the GPT-6 family: Astra, Sol and Luna.
AI-native IDE that transforms the development experience. Write code through conversation.
Discontinued: OpenAI closed the Sora app and website on April 26, 2026 and removes the Sora 2 API on September 24, 2026. Not available inside ChatGPT.
OpenAI's image generation engine. Built into ChatGPT with excellent prompt understanding.