Anthropic, the artificial intelligence company behind the Claude family of models, disabled access to its two most capable AI systems — Fable 5 and Mythos 5 — on June 12, 2026, citing a directive from the U.S. government related to export control regulations. The company confirmed the action was taken to remain in compliance with national security guidelines, marking a significant moment in the federal government's effort to regulate the international spread of frontier AI technology.

The directive, which Anthropic said it received from government authorities, required the company to restrict access to the models in question. CNBC reported that Anthropic moved quickly to disable the systems once the directive was received, with the company stating it was cooperating fully with the government's request. The models affected — Fable 5 and Mythos 5 — are understood to represent some of the most powerful AI systems Anthropic has developed to date.

The action underscores a broader push by the Trump administration to treat advanced AI models as potential national security assets subject to export licensing and control, similar to restrictions historically applied to advanced semiconductors and encryption technology. Anthropic's compliance signals that AI developers are increasingly navigating the intersection of commercial operations and federal security mandates.

According to reporting by Axios and the Associated Press, the situation is part of a wider national security framework being applied to frontier AI. Officials have expressed concern that unrestricted access to top-tier AI systems could provide strategic advantages to adversaries if the technology were accessed or replicated abroad. Anthropic has not disclosed which specific foreign access prompted the directive or whether the restriction is temporary.

The episode is expected to prompt discussion across the AI industry about how companies should respond to government directives affecting their products, and whether clearer regulatory frameworks are needed to guide both developers and federal agencies as AI capabilities continue to advance rapidly.