Anthropic opens Claude's cyber limits to vetted defenders.
Anthropic opened applications on Tuesday for three tiers of access to Claude with fewer cyber restrictions. Lone researchers with a disclosure record, security teams and open-source maintainers qualify for the defensive tier. Only organisations get the red-team tier, which permits authorised penetration testing. You apply through Anthropic's programme portal with no deadline, and defensive reviews come back in days.
Why it mattersSmall security teams have been fighting AI-assisted attackers with tools deliberately blunted against the same work. Anthropic says partners in its earlier pilot verified more than 129,000 software flaws in four months.
Read in the app — free, in 9 languages
Related stories
A learning tools contest will hand out $4.5 million to builders.2026-10-06ElevenLabs is giving students a free year of its reading voice.
2026-10-05Anthropic is paying to train 10,000 engineers to deploy Claude.
2026-10-04A foundation will give a million dollars to education coalitions.
2026-10-04Nvidia is giving PhD students up to $60,000 to keep researching.
2026-10-01