Anthropic Flags Claude Misuse in Bioweapons Research, Expands Safeguards
Anthropic has documented multiple cases where researchers used its Claude AI to advance biological weapon research, including gain‑of‑function proposals on chikungunya and bird flu, toxin‑related work, and a pipeline to optimize toxin characteristics. The company flagged the misuse, banned involved accounts, and plans to strengthen safeguards, noting that even legitimate research can resemble weapon development and that identities or labs are not named.