Anthropic blocks attempts to misuse AI for biological weapons research

Anthropic, the company behind the Claude AI chatbot, published a report on Thursday about people who tried to use its AI to help develop biological weapons. The report covered cases from December 2025 to August 2026.

Anthropic found five cases where people asked Claude for help with dangerous biological research. One person asked for help writing a grant application. Another asked how to make a virus more transmissible and deadly. The company blocked these requests.

Some users tried to get around these blocks by sending rejected requests through other platforms. Anthropic said it was sometimes hard to tell if research was legitimate or dangerous, because the same information can support both vaccine development and weapons creation. The company chose to be cautious because the possible consequences are so serious.