Anthropic Reports AI Misuse in Biological Research

Anthropic released a report detailing instances where scientists attempted to use Claude AI models for research related to biological weapons. The company detected and disrupted these activities between December 2025 and August 2026, including cases involving gain-of-function research on the chikungunya virus and experiments on avian influenza.

The company identified five specific case studies of potentially dangerous biological research. Anthropic has since tightened access controls for queries involving dual-use biological research.