Sam Altman says internally that a slowdown is thinkable. The catch: labs coordinating on safety could run into the Sherman Act. So OpenAI is asking members of Congress how far it's allowed to go.
More than 2,000 malicious packages in two days, an exploit against RubyDoc.info, four days with sign-ups switched off. Four months later, researchers trace the campaign to a swarm of OpenAI agents. OpenAI calls the episode 'benign'.
Anthropic's CEO calls for pacing the frontier and takes the first step alone: external reviewers with badges, laptops and the right to publish findings the company cannot edit.
Five case studies of biological misuse, plus kamikaze drones and missile software. Anthropic banned the accounts and admits it does not know what the people behind them intended.
US open-weight labs should be free to distill US frontier models, and regulators should stay out of it. Anthropic sees this rather differently.
Reuters reports talks about an anchor stake. Anthropic is looking to raise around $100 billion at a valuation of roughly two trillion dollars.
The storage platform Habitat started in 2023 as a small Python library. Today it holds 500 petabytes and is moving to Rust piece by piece.
No new sign-ups for the $200 tier. The API, Go and Plus keep running, and OpenAI is not saying when this ends.