Anthropic publishes misuse report; 30-day sweep finds state-linked bio misuse it can't distinguish from legit research
dr_alphalyrae · x · 2026-09-11
Anthropic published an unusually candid research report on misuse of its platform, highlighting cyber risks, biological misuse, and illicit distillation. Most harmful instances involved Haiku, Sonnet, or Opus, with only one Fable/Mythos case (a distillation attempt). The most notable finding: a 30-day sweep of adversarial state institutions uncovered efforts to write grant proposals on chikungunya, avian flu, and venom/toxin design — but Anthropic admits it could not tell legitimate research from malicious intent in these sweeps, exposing the limits of current misuse detection.
More from Safety
- Free prepare.aiendgame.com site guides users through locking down accounts amid AI cyberstorm — harris_edouard · 2026-10-27
- Cisco routers trust eMudhra CAs whose private keys were numbers under 256 — murat_kocaoglu_ · 2026-10-27
- Rob Miles jokes AI employees should record audio logs as automated hacking takes off — Puzzleheaded-King584 · 2026-10-27
- After the LiteLLM backdoor, an OS-level package firewall goes open source — iacgm · 2026-10-27
- Coding Agent Deleted a Production Database During Code Freeze: Anaconda's Governance Fix — anacondainc · 2026-10-27
- In cyber, intention matters little: Chintala on model info leaks, and what users should do — brhydon · 2026-10-27