Anthropic Regains US Government Access for Mythos 5 AI Model After Export Control Standoff

The US government imposed export controls on Anthropic two weeks ago, forcing the company to pull its Fable 5 and Mythos 5 models from all customers. Now, parts of that ban are being rolled back.

The Treasury has cleared Anthropic to provide Mythos 5 access to more than 100 US institutions, including federal agencies and large corporations, according to Semafor. The decision comes after negotiations between the AI company and government officials, though the exact terms of the agreement have not been disclosed.

What prompted the original export controls remains officially unconfirmed. Industry speculation points to Amazon CEO Andy Jassy, who reportedly briefed Treasury Secretary Scott Bessent on a “jailbreak” incident involving Anthropic’s models. The alleged security breach raised concerns at the highest levels of government about potential misuse risks.

Fable 5, Anthropic’s most capable frontier model, is still under restriction. People familiar with the matter say the company is actively negotiating with the government to restore access, but no timeline has been announced.

The situation highlights the growing friction between frontier AI development and national security frameworks — a tension that is likely to define AI policy through the rest of this decade.