Anthropic Report Details AI Misuse for Bioweapons, Hacking

Anthropic has disclosed that it disrupted multiple attempts to misuse its Claude AI models for biological weapons research, cyber espionage, and conventional weapons development, according to a threat intelligence report published on September 10, 2026. The company detailed five cases where scientists used its models for research with potential dual-use applications, including gain-of-function work on the chikungunya virus at a military institute and studies on avian influenza mammalian adaptation. Anthropic stated it banned the accounts involved but could not definitively determine if the research was intended for malicious purposes. The company withheld the identities of the researchers, their institutions, and the countries involved, citing the dual-use nature of the work. Jacob Klein, Anthropic's head of threat intelligence, noted that while the intent was unclear, a military institution conducting gain-of-function research is concerning. The report also documented a suspected Russian-linked cyberespionage campaign attributed to the group Midnight Blizzard. This operation targeted over 20 Ukrainian government, military, and defense organizations.