The Commerce Department on Friday invoked national security export controls to bar Anthropic PBC from distributing Claude Fable 5 and Mythos 5 to any foreign national, a directive that arrived at the company at 5:21 p.m. ET and forced both frontier models offline worldwide within 90 minutes. Foreign nationals couldn’t be reliably segregated from the user base, so the shutdown swept the entire customer roster. Access to Anthropic’s other models was unaffected.
The trigger came from a competitor. Andy Jassy, chief executive of Amazon, told the White House that Amazon researchers had used a series of prompts to elicit restricted information about cyberattacks from the Mythos-class system. An Amazon spokesperson told Fortune it was “not uncommon for governments to seek our counsel on potential security risks.”
What Anthropic characterized as a narrow bypass, the administration treated as cause for a recall. In calls with senior officials, CEO Dario Amodei argued the finding wasn’t a universal jailbreak. David Sacks, the White House AI adviser, posted on X over the weekend that the administration had asked Amodei to fix or withdraw the model, and that the export control followed his refusal. Semafor, citing unnamed sources, reported that U.S. officials also suspected a Chinese-linked group had accessed Mythos. An Anthropic spokesperson said that concern was never raised with the company.
The structural read is harder to miss. A rival hyperscaler walked a safety finding into the West Wing, and an export-control regime built for chip fabs and missile components was repurposed, inside an afternoon, against a commercial model. Anthropic is already contesting a Department of Defense supply chain risk designation in court while pursuing a public listing. It says it’s working to restore access. The precedent, by then, is set.
Sources
- https://www.anthropic.com/news/fable-mythos-access
- https://fortune.com/2026/06/14/how-a-warning-from-amazon-led-the-white-house-to-shut-down-anthropics-mythos-model/
- https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/
- https://www.bloomberg.com/news/articles/2026-06-14/carney-says-anthropic-ban-shows-risk-of-relying-on-big-ai-models
- https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html