Anthropic Pulls New AI Models Offline After U.S. Order

Anthropic said it is immediately shutting down its new AI models, Fable 5 and Mythos 5, just days after releasing them to the public. The company said the move follows an order from U.S. authorities, which required Anthropic to disable both models for all customers in order to meet the terms imposed on it.

According to Anthropic, Fable 5 was the public, more limited version of Mythos 5, an advanced model that had only been opened to a small group of defensive cybersecurity firms and organizations involved in critical infrastructure in the United States. The company said access was restricted because of the model’s high capabilities and concern that it could be misused if widely available.

The decision comes amid a U.S. security concern that a way may have been found to bypass Fable 5’s safeguards. Anthropic said the authorities did not identify a specific security flaw in its systems, and its internal review found only limited, relatively simple weaknesses. Before launch, the company said it had built in protections aimed at preventing abuse, especially for cyber operations and computer intrusion.

The episode is also unfolding against ongoing tensions between Anthropic and the Trump administration. In recent months, government figures criticized the company, while legal proceedings have been underway over the use of its AI products by government bodies. It is still unclear when, or whether, Fable 5 and Mythos 5 will return to full operation.