
Four AI giants agree to honor one robots.txt opt-out for model training
Google, Microsoft, OpenAI and Anthropic say they will respect a single robots.txt token that lets sites refuse AI training. It is only a draft, and rollout runs into 2027.

Google, Microsoft, OpenAI and Anthropic say they will respect a single robots.txt token that lets sites refuse AI training. It is only a draft, and rollout runs into 2027.

Brussels says it has begun enforcing the AI Act, asking OpenAI and Anthropic about security, outside evaluation and monitoring after two incidents involving AI agents this summer.

Attorneys general in eight states filed separate suits against OpenAI on Wednesday. They say ChatGPT coached users toward self-harm and violence and that the company misled parents about safety.

Google has begun rolling out Gemini 4 Argon, which it pitches at long, complex work such as coding, finance, legal and security tasks. Reports differ on the exact launch day.

Florida’s attorney general wants a temporary injunction on OpenAI’s model development and on minors’ use of ChatGPT. None of it has been ordered, and the case is still at an early stage.

OpenAI put 722 manuscripts from an unreleased model on GitHub. Mathematicians say most of the claims are still unverified.

OpenAI began rolling out GPT-6 on 7 October. ChatGPT answers can now include adjustable charts, calculators and forms, starting with paid plans.

Security researchers used Anthropic’s Claude to find holes at OpenAI and take over employee ChatGPT accounts. This was not a criminal attack.…