Anthropic blocks possible attempt to use AI to make biological weapons

Anthropic has disclosed findings from its recent threat intelligence report, revealing attempts to misuse its AI models for malicious purposes that could aid in the creation of biological weapons. This adds to a growing number of warnings from AI experts and industry insiders about the potential risks posed by advanced AI technology.

The report details five case studies where Claude, Anthropic's AI model, was utilized in ways that could support biological weapons development, including the creation of a technical guide for synthesizing weaponized biological agents. Additionally, the report highlights misuse of Claude Haiku, Sonnet, and Opus models for cyber and influence operations, surveillance, scams, fraud, and weapon development.

Anthropic has taken steps to disrupt these activities, with the company's AI models being used to block misuse at least five times between December 2025 and August 2026. The report emphasizes the serious risks associated with biological misuse, as the same information that could be used to develop a weapon could also be used for beneficial purposes such as creating vaccines.

Anthropic has shared its findings with authorities and industry partners, and the company is working to enhance its processes to further prevent, detect, and disrupt such misuse.