Amazon CEO Andy Jassy personally raised security concerns with U.S. Treasury Secretary Scott Bessent and other government officials, triggering an export control ban on two of Anthropic’s most powerful AI models, according to multiple reports. The unprecedented government action has cut off worldwide access to Anthropic’s Claude Fable 5 and Mythos 5 models, escalating a confrontation that pits the tech giant against a company in which it holds a major investment stake.
Amazon Researchers Flagged Cyberattack-Ready Capabilities in Claude Fable 5
Jassy told Bessent and other officials that Amazon researchers had successfully used Anthropic’s Claude Fable 5 to obtain information that could be directly applied in cyberattacks. The disclosure prompted the U.S. government to impose an export control ban on both the Fable 5 and Mythos 5 models effectively immediately.
An Amazon spokesperson declined to share details of the discussions but acknowledged that it is “not uncommon for governments to seek our counsel on potential security risks.” The spokesperson also pointed to an AWS health status update confirming that the company’s cloud infrastructure has been affected by the model cut-off.
Dual Role: Amazon as Investor and Competitor Raises Questions
Amazon is a major investor in Anthropic, having committed billions in funding. The development puts the company in an unusual position: as both a financial backer and the party that alerted regulators to security flaws in Anthropic’s technology. Reports from multiple outlets confirm that Amazon communicated concerns about the security of Anthropic’s models before the government moved to restrict access.
David Sacks Describes a Jailbreak That Anthropic Refused to Fix
David Sacks, who served as Trump’s AI czar and now co-chairs the President’s Council of Advisors on Science and Technology, offered a detailed account of the events. Sacks claimed that “a highly credible trusted partner of both Anthropic and the USG came forward with a jailbreak.” According to Sacks, the administration asked Anthropic CEO Dario Amodei to fix the jailbreak or de-deploy the model. “Dario refused,” Sacks stated.
Anthropic Defends Its Models, Points to Widely Available Capabilities
Anthropic responded in a blog post arguing that the capabilities that have drawn government scrutiny are already available in other publicly accessible models. The company did not elaborate on which specific capabilities it was referring to, but the statement suggests Anthropic believes the government’s action is misdirected or disproportionate.
What This Means for Security Teams and Anthropic Users
The incident represents a significant escalation in the regulatory oversight of frontier AI models. Export controls of this nature are typically reserved for sensitive military or dual-use technologies. The fact that a major cloud provider and investor flagged the vulnerability directly to the Treasury Secretary underscores the gravity of the concerns.
For organizations using Anthropic models, this situation creates immediate uncertainty. Access to the banned models has been terminated worldwide, and AWS customers have already reported disruptions. The broader implication is that AI providers may face increasingly aggressive government intervention when models are perceived to pose national security risks.
How to Assess Your Exposure and Respond
If your organization relies on Anthropic’s Claude Fable 5 or Mythos 5 models, you should immediately verify whether your access has been affected and identify alternative models or providers. Audit any workflows, automation, or security pipelines that depend on these models. This situation also highlights the importance of maintaining fallback options and avoiding lock-in to any single AI provider. Businesses should evaluate their AI supply chain for concentration risk and ensure they have documented contingency plans for sudden changes in model availability.
