Anthropic Secures Approval from US Government for Relaunch of Mythos Cybersecurity AI Model
Following a directive from the US government, access to Mythos and Fable has been halted.
Anthropic has received US government approval to reintroduce Mythos 5 to a selection of domestic organizations responsible for maintaining critical infrastructure. Although the precise number of organizations gaining access has not been disclosed, reports indicate that this includes over 100 institutions, such as large corporations and government bodies.
The company is working swiftly to reinstate access to Mythos 5 and is collaborating with federal authorities to broaden availability further. Additionally, discussions are ongoing regarding the reinstatement of Fable 5, though no specific timeline has been provided.
Previously, on June 12, Anthropic had suspended access to both Mythos 5 and Fable 5, which is intended to extend some functionalities of Mythos to a wider audience. At that time, the suspension was reportedly due to government concerns about foreign nationals accessing the models, even those physically located in the US and working for Anthropic. The government’s decision was influenced by warnings from Amazon and other organizations about potential exploitation of the models by malicious entities. There were also fears regarding a group linked to China gaining access to Mythos 5.
According to reports, Commerce Secretary Howard Lutnick communicated his decision in a letter to Anthropic, stating that adequate safeguards were in place to allow selected trusted partners to access the Claude Mythos 5 Model. He acknowledged the company’s commitment to collaborate with the government on establishing protocols and standards moving forward, noting that Anthropic had made considerable progress in lifting the access ban through daily communications over the preceding weeks.
Since June 12, we’ve been working closely with the US government to restore access to Claude Mythos 5 and Fable 5. Today, the government notified us that Mythos 5, our strongest cybersecurity model, can be redeployed to a set of US organizations that operate and defend critical…
— Anthropic (@AnthropicAI) June 27, 2026
