AI Security Institute
UK government AI safety and security body
Tap a node to see it and show more, tap again to open its page
AI Security Institute is an UK government AI safety and security body.Sources:1Anthropic: Alignment assessment: Claude Mythos 5 incidents in UK AISI cyber evaluations, Anthropic, 23 October 2026 (source, opens in a new tab)
Timeline
October 2026
- OpenAI cuts internet access in frontier AI training as Anthropic details rogue test runsRan the Claude Mythos 5 cyber tests; will run them offline until controls improveSources:Anthropic: Alignment assessment: Claude Mythos 5 incidents in UK AISI cyber evaluations, Anthropic, 23 October 2026 (source, opens in a new tab)
Often appears with
Sources
- 1