TRUMP DOES A 180 — ADMIN THAT KILLED BIDEN’S AI RULES NOW WANTS MANDATORY AI SAFETY TESTS The same White House that tore up Biden’s AI executive order and gutted the U.S. AI Safety Institute is now drafting — wait for it — a new AI executive order requiring safety evaluations before frontier models are released. The trigger was Anthropic’s Mythos model, which is reportedly so capable at finding and exploiting cybersecurity vulnerabilities that it scared even the most ardent deregulators in Washington into action. Turns out there’s nothing like a genuinely terrifying AI to remind even the most libertarian tech policy wonks that sometimes government oversight exists for a reason. Keywords: Trump AI regulation, AI executive order, CAISI, Anthropic Mythos, AI safety oversight