White House imposed export limits on Anthropic AI models over access concerns
The White House placed export restrictions on Anthropic's Mythos and Fable models after reports indicated a China-linked group may have gained access. A Semafor report cited national security concerns, including potential reverse engineering through distillation.
The VergeThe White House imposed export restrictions on Anthropic's Mythos AI model after receiving reports that a China-linked group may have accessed the system, according to a Semafor article published June 14, 2026. The restrictions also cover the related Fable model.
Officials expressed concern that access could allow reverse engineering through distillation, a process where a secondary AI is trained to replicate the behavior of a more advanced model.
Background on the reported breach Anthropic has previously stated that Mythos is too powerful for public release. The company said a Discord group obtained access to the model for two weeks before the breach was discovered and access was terminated. A spokesperson for Anthropic told Semafor that government officials did not mention China during discussions about the export controls.
Statements from administration figures A post on X by Trump advisor David Sacks addressed the restrictions but did not reference China. The post instead highlighted reported jailbreak vulnerabilities in Fable and Mythos, which Anthropic has denied. The White House has not confirmed the Semafor report.
