Amazon CEO Andy Jassy Sparked First Frontier AI Export Ban

A single phone call set off the first U.S. government action to forcibly pull a commercial frontier AI model from the market. According to The Wall Street Journal, Amazon CEO Andy Jassy spoke with Treasury Secretary Scott Bessent on June 11 and told him that Amazon researchers had used a series of prompts to get Anthropic's Fable 5 model to provide information that could be used to aid cyberattacks — output that was supposed to be off limits. The researchers had documented what they called a jailbreak that allowed users to bypass safety rails. Within days of Fable 5's launch, Commerce Secretary Howard Lutnick hit Anthropic with unprecedented export controls, forcing the lab to pull its most powerful models. Anthropic cut off access to Fable 5 and Mythos 5 for every customer worldwide on June 12, having received a U.S. government export-control directive at 5:21 PM ET that day.

The move marked the first time the U.S. government explicitly limited the release of a frontier AI model. David Sacks, who co-chairs the President's Council of Advisors on Science and Technology, said publicly that a trusted partner had presented evidence of a security vulnerability and that Anthropic CEO Dario Amodei refused to fix it. Amazon, for its part, said through a spokesperson that it is "not uncommon for governments to seek our counsel on potential security risks" but that it does not "share the details of those discussions."

Anthropics pushed back directly. The company argued that the vulnerabilities flagged by Amazon are relatively basic, that other publicly available models are capable of producing the same outputs, and that the finding does not represent a full jailbreak. Anthropic said it disagreed "that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people." By week's end, despite sending technical staff to Washington, the company remained at an impasse with the government.

What the reporting does not resolve is the structural tension at the center of the story: Amazon is both Anthropic's largest cloud infrastructure partner and, through AWS Bedrock, a competitor offering rival frontier models. Whether that dual role shaped how the vulnerability was framed and escalated to officials is not addressed in available sources. For developers and enterprises that built products on Fable 5 or Mythos 5, the most concrete lesson is operational: model availability is now a geopolitical risk surface, and the export-control mechanism deployed here could be triggered again with similar speed.