Anthropic: China, Russia actors weaponize Claude AI for bioweapons research

Anthropic reports that state-linked actors in China and Russia have circumvented safeguards on its Claude AI to support dangerous activities, including possible bioweapons research. The claim is based on Anthropic's internal threat intelligence, with limited independent corroboration. This marks a notable escalation in AI-enabled security threats, raising concerns about dual-use AI capabilities and export controls.