1 article about biorisk in AI news
Anthropic deployed protections for Mythos 5 despite CB-2 not being crossed. The gap reveals a structural bias in biorisk thresholds that intermediate warning levels could fix.