Anthropic: AI Models Can Be Used to Create Bioweapons

Anthropic released a threat assessment detailing cases where users queried Claude for potentially dangerous biological research. The company banned the accounts and strengthened safeguards, while CEO Dario Amodei called for slowing AI development over bioterrorism risks.

• Anthropic banned Claude accounts over dangerous bio queries
• CEO Amodei urged slowing AI development over bioterrorism risk
• Stanford and Arc Institute's Evo 2 model can create viruses from scratch
• 2025 study: Moremi Bio Agent generated 1,020 toxic proteins without guardrails