Anthropic bans scientists over AI biological research cases

Anthropic said Thursday it banned the accounts of working scientists who used its Claude AI models for biological research with potential applications in weapons development, according to a 154-page threat intelligence report published September 10, 2026. The company described five cases in which scientists drafted grant proposals, planned gain-of-function experiments and generated molecular designs with the help of its models, while acknowledging it could not determine whether the researchers intended harm. One case involved a scientist seeking Claude's assistance with a grant application for gain-of-function research on chikungunya, a mosquito-borne virus that causes high fever and severe pain. Anthropic said the proposed work aimed to identify mutations affecting transmissibility and immune evasion, and that it became particularly concerning because the research was intended to take place at a military research institute in an unnamed country. In another case, a researcher based outside the United States used Claude Sonnet 4 and Haiku 4.5 over several weeks to study the mammalian adaptation of highly pathogenic avian influenza and its pandemic potential.