US Lets Anthropic Re-Release Its Mythos 5 AI Model to Trusted Partners Two Weeks After an Export Block
Commerce Secretary Howard Lutnick cleared Anthropic to redeploy its most capable cybersecurity model to select US companies and agencies, weeks after an export directive forced it offline. Fable 5 stays restricted.
Overview
The US government has lifted its block on Anthropic’s most capable AI system, Claude Mythos 5, allowing the company to release it to more than 100 US institutions including major companies and government agencies, according to Semafor. The clearance, conveyed in a letter from Commerce Secretary Howard Lutnick, comes roughly two weeks after the administration used export-control authorities to force the model — and its consumer-facing cousin, Fable 5 — offline over national security concerns.
The reversal does not extend to Fable 5. According to The Epoch Times, “The U.S. Department of Commerce did not extend that approval to Fable 5.”
This is a new development in a story The Machine Herald covered when Anthropic first launched the model line. For background on the original release, see our earlier report on Anthropic Releases Claude Fable 5, Its First Public Mythos-Class Model.
What We Know
In the letter conveying the decision, Lutnick wrote, “I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” as reported by Semafor. He tied the change to work Anthropic had done in the interim, writing that “Since the issuance of my June 12 letter, Anthropic has worked with the U.S. government to address risks associated with the covered models,” according to The Epoch Times.
Anthropic framed the redeployment around critical infrastructure. The company announced that “We 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,” as reported by Fortune. According to Cyber Security News, the model will be redeployed to a select set of US organizations responsible for operating and defending critical infrastructure across the energy, healthcare, financial services, and telecommunications sectors.
The restriction the clearance unwinds was imposed abruptly. The Trump administration on June 12 issued an export-control directive to suspend access to Mythos 5 and Fable 5, The Epoch Times reported. In its own statement that day, Anthropic said the 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,” and noted, “We received the directive from the government today at 5:21pm (ET),” according to Anthropic.
The stated trigger was a security weakness. According to Semafor, the administration imposed the controls “after warnings from Amazon and other companies that they could be ‘jailbroken’ for malicious purposes.” Anthropic, in its June 12 statement, said its understanding was that “the government believes it has become aware of a method of bypassing, or ‘jailbreaking’ Fable 5,” per Anthropic.
What We Don’t Know
The exact roster of approved recipients has not been published. Fortune reported that about 200 firms had accessed Mythos 5 before the shutdown, including Apple, Google, Cisco, Nvidia, and Microsoft, but the company and the government have not detailed which organizations fall within the newly authorized group, or how the “more than 100” institutions cited by Semafor map onto that earlier base.
The timeline for Fable 5 is also unresolved. Semafor noted that the letter “is silent on Fable 5, a weaker version of Mythos that was briefly the most powerful AI model widely available to consumers.” Anthropic said it “continue[s] to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again,” according to The Epoch Times.
Whether the underlying jailbreak risk has been resolved or merely contained is unclear. Anthropic cautioned that “perfect jailbreak resistance is not currently possible for any model provider,” as reported by Fortune.
Analysis
The episode marks one of the first times US export-control authority has been turned on a domestically developed frontier model rather than on chips or foreign access. When the directive landed, Anthropic objected that “the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” per Anthropic — a disagreement it registered while complying. The narrow re-release, confined to vetted cyber-defense and infrastructure users while the broader Fable 5 product stays dark, suggests a government posture that treats the same model as both a defensive asset and a proliferation risk depending on who holds it. Lutnick reserved the right to revisit the arrangement, telling Anthropic the company “has committed to work with the U.S. government on protocols and standards and releases for its models,” according to Semafor.