Amazon CEO reportedly raised concerns before Anthropic export ban

The US government has imposed an export control ban—defined as a government restriction on the transfer of technology—on two Anthropic AI models, Claude Fable 5 and Mythos 5. The regulatory action follows reports that Amazon CEO Andy Jassy may have been the source of the security concerns that led Anthropic to cut off worldwide access to the two models on Friday. According to reports from The Wall Street Journal, Jassy communicated directly with US Treasury Secretary Scott Bessent and other government officials, stating that Amazon researchers had used Anthropic’s Claude Fable 5 model to obtain information that could be used in cyberattacks. Reports from Reuters and The Information similarly indicated that Amazon, which is a major investor in Anthropic, had communicated concerns about the security of Anthropic’s models.

The regulatory dispute escalated after David Sacks, Trump’s former AI czar who now co-chairs the President’s Council of Advisors on Science and Technology, offered his own account of the discussions. Sacks asserted that a highly credible trusted partner of both Anthropic and the US government came forward with a jailbreak, which is a technique used to bypass AI safety filters. According to Sacks, the administration intervened directly: “The Admin asked [Anthropic CEO Dario Amodei] to fix the jailbreak or de-deploy the model. Dario refused.”

Amazon confirmed that its cloud platform, AWS, has been affected by the model cutoff, pointing to an update on the service’s status. An Amazon spokesperson stated that while it is not uncommon for governments to seek the company’s counsel on potential security risks, Amazon does not share the details of those discussions. Meanwhile, Anthropic defended its position in a blog post, asserting that the capabilities causing government concern are already available in other publicly accessible models.

Why it matters

The export ban highlights the growing tension between AI developers and government regulators over the dual-use nature of AI models, specifically regarding how tech investors influence safety policy.