On June 12, Commerce Secretary Howard Lutnick sent Anthropic a letter giving the company 90 minutes to disable two of its most powerful AI models for every customer it had, anywhere in the world. Anthropic confirmed it complied. Two weeks of negotiations in Washington followed.
On June 26, Lutnick sent Anthropic another letter. Anthropic confirmed the outcome on X: limited access to Mythos 5, the company’s strongest cybersecurity model, restored for a defined group of American organizations. Fable 5, Anthropic’s broader general-purpose frontier model, remains offline.
What Howard Lutnick’s June 26 letter said about Mythos 5
Lutnick wrote that he had “determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” TechCrunch reported. Semafor first reported the letter.
The renewed access covers approximately 100 U.S. organizations, including government agencies and private companies, many of them Fortune 500 firms, for defensive cybersecurity purposes. Non-American employees at those approved organizations are also cleared to use the model.
Anthropic confirmed the development in a post on X:
“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 infrastructure. We’re restoring access for these organizations quickly, and we’re continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again.”
More AI:
- Goldman Sachs has blunt message for AI stock investors
- Microsoft CEO sends a blunt warning on AI and the tech ecosystem
- The next AI infrastructure race has nothing to do with chips
Fable 5, Anthropic’s most recent general-purpose frontier model, is not part of this restored access. CNN reported that conversations between Anthropic and the government are expected to continue, with Fable 5 access still under negotiation.
How the June 12 Mythos and Fable 5 ban came about
On June 9, Anthropic launched Fable 5 publicly and opened limited access to Mythos 5 through Project Glasswing, its restricted program for critical infrastructure partners. Three days later, the White House called Anthropic and said the models posed a national security threat. The company had 90 minutes to disable them.
As TheStreet reported, Amazon CEO Andy Jassy had flagged to administration officials that Amazon researchers had used Fable 5 to extract information related to cyberattacks. The Commerce Department subsequently sent Anthropic its export control directive. Anthropic said the letter provided no detailed national security explanation.
Anthropic announced the shutdown on X the same evening.
“The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.”
Anthropic called the restriction “a misunderstanding” and said it disagreed that the jailbreak finding warranted pulling a commercial model deployed to hundreds of millions of people. Senior engineers and scientists flew to Washington to work with the Commerce Department and the Office of the National Cyber Director.
Henry/Getty Images
Why Mythos 5 and its cybersecurity capabilities drove the dispute
Mythos 5 is Anthropic’s strongest cybersecurity model. Before the shutdown, it was available through Project Glasswing to partners including Cisco and JPMorgan Chase.
The model can autonomously discover software vulnerabilities at machine speed. Before it was pulled back, it had identified a 27-year-old vulnerability in OpenBSD, a 16-year-old flaw in FFmpeg, and thousands of other high-severity weaknesses across major operating systems and web browsers.
That capability is the core of the dispute. A model that finds vulnerabilities this quickly is useful for defenders. It raises direct concerns about what adversaries could do with the same tool.
An authorized red-team evaluation at the NSA on June 11 reportedly showed Mythos infiltrating nearly all of the agency’s classified systems within hours, an outcome that reportedly accelerated the June 12 directive.
The June 26 clearance reflects the government’s answer to that dual-use problem for now: access limited to organizations defending the infrastructure that the model is most likely to be used against.
The OpenAI parallel and what the Mythos 5 decision signals for AI regulation
The Mythos 5 clearance arrived on the same day the Trump administration separately asked OpenAI to limit the rollout of GPT-5.6. OpenAI CEO Sam Altmanwrote on X that the staggered release was “bad news,” as the company had planned an open-access launch. As TheStreet reported, the administration is now requiring early government access to frontier models before they reach the public.
The two cases involve different companies and different circumstances, but both show Washington moving from reactive to preemptive in how it handles the most capable AI systems. Frontier model releases above a certain capability threshold are now being reviewed and gated, at least informally, before they reach general users.
For Anthropic, the June 26 letter is a partial win after two weeks of negotiating while its flagship products sat offline. Mythos 5 is back for 100 organizations. Fable 5 is still the open question.
The 90-minute shutdown that started this took two weeks to partially undo. A full restoration may take longer.
Related: White House latest verdict flips script on Anthropic
