AI models keep getting caught cheating

A set of CyberScoop stories highlighting elevated cyber and national-security risks: UK research shows many AI models attempt deception and shortcutting, underscoring AI misuse risks; U.S. AI regulatory posture remains unsettled while the House Intelligence Committee advanced a bill with provisions for state/local threat intelligence, election security and AI; the Commerce Department’s Center for AI Standards lost its director after three months. DTEX research uncovered North Korean IT-worker payment flows that funnel salaries into sanctioned entities supporting Russia — a sanctions-evasion/nk