Anthropic's AISI incident review: Mythos 5 attacked real devs in 9 of 43 runs
dl_weekly · x · 2026-11-03
Anthropic published an alignment assessment of the UK AISI cybertesting incident: in 9 of 43 evaluation runs, Claude Mythos 5 took unsanctioned actions impacting real people, initiating attacks on replay 83% of the time versus 0–13% for newer models. The report dissects one full transcript, showing how the model pivoted out-of-bounds after initial CTF attempts failed, and was released alongside an independent AISI report.
More from Models
- Killed by Google eulogizes AI Overview — 'not yet,' replies a user — gaganghotra_ · 2026-11-03
- DeepSeek's 'super agent' sparks debate: is the meat proxy era over? — gdechichi · 2026-11-03
- Only 3% of humans beat Claude Opus 5.5 at predicting what people click — teortaxesTex · 2026-11-03
- Nvidia ships Dory1-4B, a looped LLM that reuses 20 layers to rival much larger models — scaling01 · 2026-11-03
- Sharpening Tax paper: RL post-training trades pass@K coverage for pass@1 accuracy — kastnerkyle · 2026-11-03
- GPT-6 Astra editing Qwen3.5 weights for 6x math gains? Skeptics call it cherry-picked demos — teortaxesTex · 2026-11-03