Anthropic Blocks AI Weapons Misuse

Anthropic said in a threat intelligence report released Sept. 10 that it banned accounts in five cases where working scientists used Claude for biological research that could support weapons development. The company said users bypassed controls or obscured their aims, including a gain-of-function proposal involving a mosquito-borne virus, but said it could not determine intent. The report also said an Iran-linked group used Claude to compile targeting information on U.S. Navy ships, and users in Houthi-held Yemen sought help developing missile software. It also cited attempted cyberattacks, surveillance and other weapons-related assistance.