Amazon Chief Executive Andy Jassy reportedly triggered United States export restrictions on two artificial intelligence systems after raising security allegations with federal officials. According to The Wall Street Journal, Jassy informed Treasury Secretary Scott Bessent and other agency representatives last Friday that Amazon researchers utilized Anthropic’s Claude Fable 5 to extract data applicable to cyber operations. These disclosures prompted the government to implement an immediate export control ban on the Fable 5 and Mythos 5 architectures, leading Anthropic to suspend global access to both systems. Amazon acknowledged receiving government inquiries regarding technology security but declined to elaborate on specific conversations. A company representative noted that consulting with federal agencies is standard practice and that internal discussions remain confidential. Parallel reports from The Information and Reuters corroborated that Amazon, which holds a significant financial stake in Anthropic, independently communicated safety apprehensions to regulators. David Sacks, co-chair of the President’s Council of Advisors on Science and Technology and former White House artificial intelligence coordinator, provided an alternative account of the regulatory intervention. Sacks stated that a trusted intermediary representing both Anthropic and the United States government disclosed a critical security vulnerability involving a model jailbreak. He asserted that federal authorities instructed Anthropic chief executive Dario Amodei to patch the vulnerability or withdraw the system from deployment. According to Sacks, Amodei declined to implement the requested safeguards or remove the affected models. The incident highlights intensifying scrutiny surrounding large language model safety and the escalating coordination between private technology firms and federal regulators. The export controls restrict the dissemination of the flagged systems internationally, aiming to prevent potential misuse in cyber operations. Anthropic’s decision to disable worldwide access underscores the operational adjustments required when models face unmet security thresholds. As artificial intelligence capabilities expand rapidly, these regulatory actions signal a shift toward proactive federal oversight and stricter compliance expectations for foundational model developers.