US Partially Lifts Export Controls on Claude Mythos 5

The US Commerce Department has authorized Anthropic to release Claude Mythos 5 to over 100 US institutions, partially lifting export controls imposed two weeks ago that disabled access to both Claude Mythos 5 and Claude Fable 5 for all customers.

The Export Control Directive

According to Semafor, the US government issued the original export control directive after Amazon CEO Andy Jassy reportedly informed Treasury Secretary Scott Bessent about a purported jailbreak vulnerability in Anthropic's models. Additional reports indicate the White House had grown concerned about potential Chinese access to Claude Mythos.

The directive forced Anthropic to immediately disable customer access to both Claude Mythos 5 and Claude Fable 5, affecting the company's entire customer base.

Resolution Details

In a letter to Anthropic's chief compute officer Tom Brown dated Friday, Commerce Secretary Howard Lutnick wrote: "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model." The letter cited "significant progress" in daily talks between the government and company since the block took effect.

The partial lift allows Anthropic to release Claude Mythos 5 to more than 100 US institutions, including major companies and government agencies. The specific institutions have not been disclosed.

Status of Claude Fable 5

Lutnick's letter makes no mention of Claude Fable 5. However, sources close to the negotiations told Semafor that discussions are moving toward releasing Fable 5 as well, though no timeline has been established.

Broader Industry Impact

The development comes hours after OpenAI announced its GPT-5.6 models would initially be available only to a small group of trusted partners approved in coordination with the US government, with broader rollout planned for coming weeks.

What This Means

This marks the first instance of US export controls forcing a major AI lab to disable model access for all customers, then partially restoring it only for approved institutions. The incident establishes a precedent where frontier AI models may require explicit government approval before deployment, even to domestic customers. The two-week negotiation period and ongoing restrictions on Fable 5 suggest the government is implementing new oversight mechanisms for advanced AI systems, potentially based on specific capability thresholds or security concerns. For Anthropic's broader customer base, the timeline for restored access remains unclear.