Anthropic's own staff say publicly that no Anthropic system or laptop was compromised, after security researchers blamed one for the tampered Claude Code extension that reached up to 5,000 Cursor users — and neither company will say how the attacker got publishing access.

Chainalysis says an external actor used 'offensive agentic techniques' to reach its systems and internal tools this month, exposing customer names, emails and role labels — while it traces North Korea's $6.3BN Binance theft, which it says is unrelated.

OpenAI, Google, Microsoft and Anthropic will honour a shared robots.txt 'policy token' proposed by an Anthropic standards lead at the IETF — arriving years after the scraping it governs, and the same week OpenAI published a defence of fair use.