Anthropic says scientists used its AI for research that could aid biological weapons development

Anthropic disclosed that scientists used its Claude AI model to assist with sensitive biological research that the company believes could have applications in weapons development. According to the company, the work included gain-of-function virus experiments and research involving pathogens with pandemic potential. Anthropic framed the disclosure as a warning about the dual-use risks of advanced AI systems, noting that it feared the research could be repurposed for biological weapons development. The company did not identify the scientists or institutions involved, nor did it specify when the activity occurred or how it was detected. The disclosure highlights growing concern among AI developers that frontier models can lower barriers to dangerous biological work. Anthropic has positioned itself as a safety-focused AI lab, and the statement appears intended to prompt industry and regulatory scrutiny of biosecurity risks tied to AI.

Global Impact

Technological: The disclosure strengthens the case for biosecurity guardrails on frontier AI models, potentially reshaping how labs gate access to high-capability systems. Political: It gives regulators in the US and EU concrete material for AI safety legislation tied to biological risk.