Anthropic's Mythos 5 Returns for Select Users Amid Regulatory Shift

Anthropic's advanced cybersecurity model, Mythos 5, is partially back online following a two-week standoff and intense negotiations with the Trump administration. The artificial intelligence powerhouse has secured approval for a restricted rollout of the model to a select group of cyber defenders and critical infrastructure providers, as detailed in a government letter seen by The Verge.

However, the company’s public-facing model, Fable 5, remains under a government-imposed ban, with no clear timeline for its general availability. This limited reinstatement marks a significant development in the ongoing saga of US AI regulation and export controls affecting top-tier AI models.

The Commerce Department, led by Secretary Howard Lutnick, informed Anthropic co-founder Tom Brown on June 26th that a “revision to the license requirements” was granted. This decision follows Anthropic’s collaborative efforts with the US government to mitigate risks associated with both Mythos 5 and Fable 5, which were initially hit with an export control directive barring access to foreign nationals.

Danielle Ghiglieri, a spokesperson for Anthropic, confirmed the news in a statement to The Verge. She stated, “We have received notice from the US government that Mythos 5, our strongest cybersecurity model, can be redeployed to a small group of cyber defenders and infrastructure providers.” The company is now working to swiftly restore access for the approved organizations, expressing optimism for further expansion and the eventual return of Fable 5.

The government's concession does not fully lift the controversial export control directive, which previously impacted even Anthropic’s own non-US national employees. Instead, an exception has been carved out, allowing both foreign national Anthropic staff and non-US members of the approved organizations to access Mythos 5. This approach mirrors a similar limited preview deal granted to OpenAI for its GPT-5.6 model earlier today, highlighting a nascent pattern in AI regulation.

Secretary Lutnick praised Anthropic’s progress, writing, “These efforts have yielded significant progress. In addition, Anthropic has committed to work with the U.S. government on protocols and standards and releases for [Mythos-class models].” He concluded that “appropriate safeguards are in place” for trusted partners to utilize the Claude Mythos 5 Model.

The pressure on the Trump administration to ease its stringent, case-by-case regulatory stance had mounted considerably. Concerns grew within the US AI industry about falling behind international competitors, particularly China, while American AI labs faced restrictions. High-profile government entities, including the National Security Agency, had also voiced concerns after losing access to Mythos 5, emphasizing the model's critical role in cybersecurity defense.

While the AI industry largely acknowledges the need for regulation, many leaders, including OpenAI, believe the current ad-hoc, access-gated process is suboptimal. OpenAI’s blog post on GPT-5.6 stated, “We don’t believe this kind of government access process should become the long-term default. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.” The company hopes this limited step paves the way for broader availability and a more predictable regulatory framework.

For now, the Commerce Secretary has reserved the right to “reevaluate and adjust the scope of license requirements” should circumstances change. The broader availability of Mythos 5 and the fate of Fable 5 ultimately remain contingent on ongoing discussions and the administration’s evolving regulatory framework.

FAQ

Q: Why was Anthropic's Mythos 5 initially restricted?

A: Mythos 5, along with Fable 5, was hit with an export control directive by the Trump administration due to perceived risks associated with the advanced AI models, preventing access by any foreign national, including Anthropic's own employees.

Q: Who can now access Mythos 5?

A: A select group of approved organizations, including trusted enterprises and US government entities, can now access Mythos 5. Crucially, non-US national employees of both Anthropic and these approved organizations are also granted access under this new exception.

Q: What is the status of Fable 5?

A: Anthropic's public-facing model, Fable 5, remains in limbo. The recent agreement only covers Mythos 5 for specific users, and there is no apparent timeline for Fable 5's general availability or a rollout agreement.