Trump admin lets Anthropic restore Mythos 5 to 100 US firms

The Trump administration partially reversed its two-week-old ban on Anthropic's most powerful AI models, allowing more than 100 US government agencies and companies to access Mythos 5 while leaving the consumer-facing Fable 5 in limbo.

Commerce Secretary Howard Lutnick informed Anthropic cofounder and chief compute officer Tom Brown on Friday that "appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model," according to a letter obtained by Semafor and confirmed by Reuters. The directive permits foreign national employees at approved organizations — including Anthropic's own non-American staff — to use the model, a significant softening from the June 12 order that barred any non-US citizen from accessing it.

"Anthropic has worked with the U.S. government to address risks associated with the Covered Models. These efforts have yielded significant progress," Lutnick wrote.

The original export control order, issued on a Friday evening two weeks ago, demanded Anthropic suspend access by "any foreign national" to both Mythos 5 and Fable 5 due to security concerns. Anthropic responded by taking both models entirely offline. The administration grew concerned after learning Anthropic had granted access to a South Korean telecommunications firm it believed had ties to China, WIRED reported. Amazon CEO Andy Jassy and the National Security Agency separately flagged a method for jailbreaking Fable 5's guardrails, convincing officials they needed to act.

The partial reinstatement covers only Mythos 5, the cybersecurity-focused model designed to identify software vulnerabilities. Fable 5, a consumer version with additional safeguards that was released days before the ban, was not addressed in Lutnick's directive. Anthropic said in a post on X that it is "continuing to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again." A person familiar with the matter said discussions are expected to continue over the weekend, with both sides hopeful the resolution will inform a lasting policy framework for future model releases.

The standoff has been costly for Anthropic. Before the ban, the company was seen as one of the few AI startups with a path to profitability, with Mythos-class models whose input tokens sell for double the cost of its lower-powered Opus 4.8. The models were expected to boost revenue ahead of an upcoming IPO. Anthropic also faces a $15 billion per year commitment to SpaceX for data center access, making the revenue from Mythos critical to its financial trajectory.

The broader implications extend beyond Anthropic. OpenAI's GPT-5.5 Cyber recently beat Mythos 5 on certain cybersecurity benchmarks, and the Trump administration reportedly asked OpenAI to delay GPT-5.6 over similar security concerns, with plans for the government to approve each customer individually. Dean Ball, head of the strategic futures team at OpenAI and a former White House AI adviser, said in a June 16 blog post that the Anthropic episode has shown frontier AI developers they "need an explicit green light from the government now."

The policy reversal comes after months of the Trump administration pushing to dismantle AI safeguards and regulation established under the Biden administration. Cybersecurity leaders have argued the export control approach is the wrong way to regulate AI. Katie Moussouris, founder and CEO of Luta Security, reviewed the Fable 5 vulnerability at Anthropic's request and said in a blog post that the alleged guardrail bypass was "the most valuable thing an AI model can do for defensive security" — not a security threat.

As Alex Stamos, chief product officer at Corridor, told The Verge last week: "One of America's champions is being kneecapped by the US government while we're in a race with the Chinese. It's just incredibly stupid."

This article is for informational purposes only and does not constitute investment advice.