US Government Allows Anthropic to Restore Access to Mythos for Select Organizations
The US Commerce Department has partially lifted restrictions on Anthropic's most advanced AI model, Claude Mythos 5, allowing more than 100 US organizations—including large corporations and government agencies—to regain access.
Access Restored
The US government has eased the restrictions it imposed on Anthropic’s most advanced AI model, Claude Mythos 5. Commerce Secretary Howard Lutnick sent a letter to Anthropic cofounder Tom Brown, stating that the company may now grant access to over 100 US organizations, including large corporations and government agencies, because he had determined that appropriate safeguards are in place.
However, the White House stopped short of permitting a broader rollout of the model and said nothing about the fate of Claude Fable 5, the consumer-facing version released with stronger safeguards. All other requirements from the initial directive issued on June 12 remain in effect.
Reason and Reactions
Anthropic spokesperson Eduardo Maia Silva said the company received notice that Mythos 5, its strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers. Silva expressed satisfaction with the progress and said the company was working to restore access quickly, with hopes to expand access further.
The partial reinstatement comes about two weeks after the White House sent an export control directive requiring Anthropic to limit foreign nationals’ access to Claude Mythos and Fable 5. In his latest letter, Lutnick wrote that approved organizations may now extend access to their foreign national employees, as well as Anthropic’s foreign national employees.
Policy Implications
The Trump administration became concerned about Anthropic’s rollout after learning the company shared access with a South Korean telecom firm believed to have ties to China. Following concerns from Amazon and the NSA that Claude Fable 5 could be jailbroken, the administration decided to act. Anthropic sent senior cybersecurity and AI safety staff to Washington, DC to meet with officials.
The incident has raised broader questions about US AI policy, particularly the extent to which the administration will control future model releases. OpenAI recently announced it was delaying its GPT 5.6 models in response to a request from the administration. Dean Ball, head of strategic futures at OpenAI and a former White House AI adviser, said that frontier AI developers now need an explicit green light from the government.
