The Trump administration imposed emergency export controls on Anthropic’s Fable 5 and Mythos 5 artificial intelligence models this past Friday, abruptly restricting foreign access to the systems. The decision followed a 24-hour escalation driven by national security concerns over reported vulnerabilities in the models’ safety guardrails. The crisis accelerated on Thursday when Amazon, an Anthropic investor, alerted the White House to instances where users bypassed the AI’s restrictions. Administration officials, including Treasury Secretary Scott Bessent and Cyber Director Sean Cairncross, convened an emergency meeting. By Friday morning, officials attempted to contact CEO Dario Amodei to demand voluntary removal of the models. While Amodei’s office initially disputed his availability, he joined three calls with senior officials within an hour and 15 minutes. During the negotiations, Amodei defended Anthropic’s safety architecture, characterizing the reported bypass as an isolated incident rather than a systemic jailbreak. He requested additional time and technical data to address the flaws. Administration officials, however, considered the evidence substantiated after review by the National Security Agency and viewed Anthropic’s reluctance to pause access as a significant security risk. Bessent reportedly warned Amodei that his stance constituted a poor decision. When voluntary compliance was not secured within a narrow window, the White House invoked national security authorities to enforce the export restrictions. The directive effectively disabled both models for foreign nationals and third-party customers pending remediation. Anthropic publicly criticized the action as disproportionate and lacking transparent regulatory grounding, though it confirmed immediate compliance to meet the order. The incident highlights the accelerating friction between rapid AI development and federal oversight. Despite Anthropic’s longstanding advocacy for industry regulation, White House officials expressed frustration over the company’s handling of the reported vulnerabilities. The administration maintains that security protocols must precede commercial deployment, a stance recently reinforced by executive orders requiring advance model submissions. Industry observers note the episode signals a shift toward proactive enforcement rather than voluntary guidance. While some officials, including former AI czar David Sacks, endorsed the controls as a targeted security measure, the standoff underscores ongoing challenges in aligning private AI innovation with national defense requirements. Anthropic has indicated that broader access to Fable 5 and Mythos 5 will resume upon successful verification of security patches.