A recent White House export control directive has restricted foreign access to Anthropic’s Fable 5 and Mythos 5 artificial intelligence models, following cybersecurity findings and executive communications originating from Amazon. According to a Wall Street Journal report, the policy was initiated after Amazon shared research demonstrating that targeted prompts could extract data exploitable for cyberattacks from the Fable 5 architecture. Amazon CEO Andy Jassy subsequently briefed White House officials, prompting the administration to issue the restriction. The directive immediately bars foreign nationals from accessing the specified models. This measure has created significant operational challenges for Anthropic, which employs a substantial number of internationally born researchers who are now prohibited from using their own company’s tools. Amazon has not issued a public statement regarding the report. Anthropic has formally challenged the administration’s characterization of the Amazon findings as a security jailbreak. In a public statement, the company argued that the demonstrated vulnerabilities are standard for current large language models and could be replicated using other commercially available systems, including GPT 5.5. Several independent security experts have aligned with Anthropic’s position. Katie Moussouris, founder and CEO of LutaSecurity, confirmed she reviewed the Amazon paper and concluded it does not constitute a jailbreak. Former Commerce Department official Kate Koren suggested to the Journal that preexisting administrative friction may have influenced the enforcement approach. The export restriction marks a resurgence in longstanding policy disputes between Anthropic and the Trump administration. Earlier conflicts centered on the company’s refusal to deploy its technology for domestic mass surveillance or lethal autonomous weapons systems. In February, the administration ordered federal agencies to cease Anthropic AI usage, and Defense Secretary Pete Hegseth subsequently classified the company as a supply chain risk. Following temporary negotiations that temporarily expanded access to the Mythos architecture, the two parties have entered a renewed standoff. The current directive effectively isolates international development teams from critical model access, potentially delaying ongoing research and complicating Anthropic’s global operational structure. The incident underscores growing federal scrutiny over foreign access to advanced AI infrastructure and highlights the ongoing tension between corporate security practices and national export controls.