🇺🇸 A closed-door AI demo on Capitol Hill reportedly left U.S

🇺🇸 A closed-door AI demo on Capitol Hill reportedly left U.S. lawmakers shaken. Anthropic showed members of the House Homeland Security Committee how its Claude Mythos model could identify and connect software vulnerabilities to map out a theoretical attack that could compromise financial systems and potentially drain bank accounts all inside a controlled demonstration. No real money was involved. Committee Chair Andrew Garbarino also described a separate demo where a jailbroken AI model generated a detailed plan to kidnap a lawmaker in under 30 seconds, though public reports don’t clearly confirm that Mythos was the model used. The demonstrations underscored a growing concern in Washington: as AI becomes more capable, it could also become a powerful tool for cybercriminals and hostile governments if it falls into the wrong hands. @aipost 🏴