The Firewall and the Fire: What Anthropic's Chikungunya Block Actually Proves
Anthropic published its third AI threat-activity report since March 2025, disclosing that it blocked a user's attempt to conduct gain-of-function research on chikungunya virus using its AI models. The disclosure coincided with a high-profile resignation and public risk estimates from Anthropic's own safety staff, intensifying the global debate on whether voluntary corporate safeguards are enough to govern dual-use AI.