Commerce Secretary Howard Lutnick authorized limited access to Anthropic’s cybersecurity-focused Mythos 5 model for approved organizations while broader restrictions remain in place.
The U.S. Commerce Department approved the limited release of Anthropic’s Mythos 5 AI model to select organizations, while continuing to restrict access to its consumer-focused Fable 5 model.
According to a letter from Commerce Secretary Howard Lutnick reviewed by Reuters, the authorization allows Anthropic to give access to Mythos 5 to a defined group of trusted U.S. organizations following the government’s earlier suspension of both models. Axios, which also reviewed the letter, reported the decision followed what the department described as “significant progress” by Anthropic in implementing appropriate safeguards.
According to Reuters and Axios, the letter includes five key provisions:
- Limited access to Mythos 5 is restored. Anthropic may now provide the model to approved trusted organizations under the revised authorization.
- Access is limited to approved organizations. Reuters reported that more than 100 organizations, including many Fortune 500 companies and critical infrastructure partners, are eligible under the authorization.
- Foreign-national employees may use the model. Employees of approved organizations may access Mythos 5 without requiring additional export licensing, according to Reuters.
- Commerce retains the authority to modify the decision. Reuters and Axios reported that the department reserved the right to revise or withdraw the authorization if circumstances change.
Earlier this month, the Commerce Department directed Anthropic to suspend access to Mythos 5 and Fable 5 after national security concerns emerged following a reported jailbreak involving the models.
The White House decision follows another recent example of government involvement in frontier AI deployments. On the same day, OpenAI announced it would limit the release of its new GPT-5.6 Sol model at the request of the U.S. government.
Anthropic said it is working to restore access to Mythos 5 for approved organizations while continuing discussions with the U.S. government regarding the broader availability of its frontier AI models.
