Anthropic Says It Blocked Attempts to Use AI for Potential Bioweapon Research

Anthropic said it blocked multiple attempts this year by scientists to use its technology for research that could help develop biological weapons. According to ChainCatcher, the company cited five cases in a report on malicious use of its models and said some actors bypassed controls and obscured their research goals to evade safety protections. Anthropic said it shared the cases to prompt discussion across the AI industry and governments about emerging biological risks and possible responses. One case involved a researcher from an unsupported region who used Claude over several weeks to plan experiments involving avian influenza. The company said safety filters limited that research to its weakest model and noted it could not determine whether the scientist intended to cause harm. Anthropic also said the same information needed to develop biological weapons could be used to develop vaccines. It said it banned the accounts mentioned in the report but did not disclose the researcher’s institution or the country where the incident occurred.