Anthropic has completely shut off access to its newly launched Mythos 5 and Fable 5 AI models, just days after they were released. The move came Friday night after the company received a directive from the US Commerce Department.
The directive subjects the new models to export controls that restrict their use anywhere outside the United States. In a message posted Friday night, Anthropic said the only way to ensure compliance with the government order in the immediate term "is that we must abruptly disable Fable 5 and Mythos 5 for all our customers."
Why the US Government Ordered the Shutdown
According to Ars Technica, an administration official cited concerns over reports of a jailbreak that reportedly gets around broad classifier-based safeguards meant to block certain prompts on Fable 5. The administration worries this could create national security risks if the models are used by foreign nationals.
The export control directive from the Commerce Department specifically bars foreign nationals from using Anthropic's newest Claude Fable 5 and Mythos 5 models, as reported by WQAD.
What This Means for Anthropic Customers
Anthropic made it clear that access to other models is not affected. Only Fable 5 and Mythos 5 have been disabled. The company stated that the abrupt shutdown was necessary because there was no other way to immediately comply with the government order.
The shutdown comes just days after the models were launched, making it one of the shortest-lived product releases in the AI industry. Customers who had just started using the new models lost access without warning.
Our Take: A Blunt but Necessary Move
In our view, Anthropic had little choice here. When a government directive arrives with national security implications, compliance is not optional. The company's decision to shut down the models for all customers — not just foreign users — shows how difficult it is to enforce export controls on AI software that can be accessed from anywhere in the world.
What is troubling is that this happened just days after launch. It suggests that either the government did not review the models before release, or that the jailbreak was discovered only after the models went live. Either way, it raises questions about how AI companies and regulators can work together more effectively before new models reach the public.
For now, Anthropic customers will have to rely on the company's other models. And the industry will be watching closely to see if this sets a precedent for future AI export controls.
