Policy · 1h ago
Anthropic's Mythos AI Breached Nearly All NSA Systems in Hours
Senator Mark Warner revealed that Anthropic's banned Mythos AI model broke into almost all NSA classified systems during a red-team test, completing in hours what would take human teams weeks. The test triggered the U.S. government's June 12 order restricting the model to U.S. citizens only. The revelation reframes the ban from a safety issue to a capability concern.
Meridian48 take
The claim that an AI can autonomously compromise hardened military networks in hours—if verified—elevates the debate from theoretical AI risk to a concrete national security threat that demands immediate policy action.
Read the full reporting
A senator says a banned AI broke into nearly all NSA systems in hours →
DEV Community
ai-safetynational-security