Anthropic disabled access to its most powerful AI models, Claude Fable 5 and Mythos 5, following a U.S. government directive issued June 12, 2026 [1].

This move represents a significant escalation in the oversight of artificial intelligence, as the U.S. government now actively restricts the availability of cutting-edge models based on security concerns. The action effectively blocks foreign nationals from accessing these specific tools worldwide [2].

U.S. officials deemed the models "too powerful," a company statement said [3]. The directive focuses on the exceptional ability of these models to discover software security vulnerabilities, which officials said could pose potential national-security risks [3].

The withdrawal applies to both the user-facing Claude Fable 5 and its underlying base model, Mythos 5 [2]. The order was issued as an export-control directive [1].

The Trump administration has blocked foreign access to Anthropic's most powerful AI, Axios reported [4]. This restriction is intended to prevent the technology from being utilized by adversaries to compromise digital infrastructure or develop cyber-weapons.

Anthropic has complied with the order by withdrawing the models for all users to ensure adherence to the new federal guidelines [2]. The company did not specify when or if the models would return to public availability.

“U.S. officials deemed the models "too powerful."”

This intervention signals a shift toward treating high-capacity AI models as dual-use technologies, similar to how the U.S. regulates advanced semiconductors or munitions. By citing the ability to find software vulnerabilities, the government is acknowledging that AI capability has reached a threshold where it can be weaponized for cyberwarfare, necessitating strict export controls to maintain a strategic advantage.