The Trump administration concluded talks with Anthropic on Monday without lifting export controls that were imposed last week on the company’s most advanced AI models, according to three people briefed on the matter. The controls were triggered by concerns that jailbreaking techniques could disable guardrails on Anthropic’s Claude Fable 5, potentially allowing users to access the more powerful cybersecurity capabilities of the company’s Mythos model.
The Jailbreak Concerns
The administration was first alerted to the jailbreak concerns last week when Amazon CEO Andy Jassy called Treasury secretary Scott Bessent directly about the alleged vulnerabilities, a move that played a role in spooking the administration, the people said. Alarmed White House officials tasked the NSA to help review the vulnerabilities. The NSA responded that it believed it was indeed possible to strip away Fable 5’s guardrails, prompting the administration to impose restrictions on the model.
In a blog post on Friday, Anthropic implied that the administration’s characterizations of the potential risks are overblown. Some cybersecurity researchers reiterated this position to officials on Monday, sending an open letter arguing that the export control action taken against Anthropic was unjustified.
Emergency Talks and Key Players
The emergency talks were held at the Commerce Department with government researchers from the Center for AI Standards and Innovation (CAISI) and the Office of the National Cyber Director Sean Cairncross. One person said Cairncross himself did not participate. Commerce secretary Howard Lutnick dialed in by conference call from the G7 summit in Evian, France. On Anthropic’s side, cofounder and chief compute officer Tom Brown and head of external affairs Sarah Heck led the discussions. Anthropic’s head of frontier red teaming, Logan Graham, and senior security researcher Nicholas Carlini flew to Washington, DC for the talks.
“Both parties are working quickly to get this resolved,” an Anthropic spokesperson said in a statement to WIRED. A White House spokesperson declined to comment.
Security Disconnect
At the core of the conversations is a disagreement over the severity of the jailbreaking concerns. The Commerce Department expressed a willingness to find a way to bring Fable 5 back online for consumer use, but it would likely be contingent on Anthropic fully resolving the jailbreak concerns, the person said. It’s unclear why Amazon, one of the largest investors in Anthropic, rang the alarm on Fable 5. “As a leading cloud provider that serves a large number of private and public sector customers, it’s not uncommon for governments to seek our counsel on potential security risks,” an Amazon spokesperson tells WIRED. “When they occur, we don’t share the details of these discussions.”
The talks have come at a fraught political moment for Anthropic, which was already in a prolonged fight with the Pentagon over whether its AI models could be used for certain military applications.
| Aspect | Anthropic Position | Administration Position |
|---|---|---|
| Severity of jailbreak | Concerns overblown (blog post) | Believes guardrails can be stripped (NSA confirmation) |
| Export controls | Imposed last week; Anthropic cut off access | Controls continue; not lifted |
| Resolution path | Working quickly to resolve | Lifting controls contingent on full resolution of jailbreak concerns |
Implications for Enterprise AI
For enterprise adopters of AI, the Claude Fable 5 dispute underscores the growing tension between advanced AI capabilities and government security oversight. The involvement of Amazon — a major cloud provider — highlights how enterprise customers may be caught in crossfire between model developers and regulators. CTOs and technology procurement leaders should monitor these developments closely, as export controls on AI models could affect availability of cutting-edge tools for cybersecurity, automation, and data analysis. The case also signals that even large-scale AI providers like Anthropic face significant regulatory challenges that may delay or restrict access to their most powerful models.
