US restricts Anthropic Mythos and Fable AI models over fears foreign military will use them
US Commerce Secretary Howard Lutnick said he took action against Anthropic's latest Mythos and Fable AI models because officials feared they could be deployed by military intelligence users in China, Russia or other countries of concern.
Image by Cybernews.
Follow us
US Commerce Secretary Howard Lutnick said he took action against Anthropic's latest Mythos and Fable AI models because officials feared they could be deployed by military intelligence users in China, Russia or other countries of concern.
- The US government restricted Anthropic’s latest AI models over concerns they could be used by foreign military intelligence services.
- Officials ordered a suspension of global access while negotiations continue over safety controls and safeguards.
- The Commerce Department used export control powers in what is reported as a first-time application to AI models.
- Experts and industry figures have questioned the legal basis and urged the restrictions to be lifted.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
Lutnick noted the risk in a letter sent to Anthropic CEO Dario Amodei on Friday, ordering the company to suspend export of the AI models to destinations worldwide and all foreign nationals, wherever located, according to a copy of the letter seen by Reuters on Monday.
Senior Anthropic technical staff met with officials at the Department of Commerce in Washington on Monday to negotiate a solution, a Trump administration official said.
The stakes are high, with the government seeking assurances the models cannot be used to harm the US, while Anthropic is pushing to restore access to its top-tier models after taking them offline for all users on Friday.
Lutnick has been part of the process, holding regular calls with Anthropic officials as he works toward a deal, a source familiar with the situation said. Amodei and Lutnick are both set to attend the G7 meetings in Evian-les-Bains, France, where they may speak as negotiations continue. National Cyber Director Sean Cairncross joined Monday's working-level meeting with Anthropic at the Commerce Department, the source added.
The company's technical staff have met with officials virtually every day since the Trump administration contacted the company on Friday, a person close to the company told Reuters.
After the Lutnick letter, Anthropic said it would disable access to the models globally.
The government told the company it believes there is a method of bypassing, or "jailbreaking," a safeguard that would prevent Fable 5 from being used in identifying software vulnerabilities, Anthropic said in a blog post on Friday.
The bypass found only "minor" security flaws that other publicly available models can also find, the company added. Relations between the Trump administration and Anthropic ruptured earlier this year after Anthropic refused to allow the US military to use its AI models for domestic surveillance and fully autonomous weapons systems, and the government retaliated by putting it on a national security blacklist.
The San Francisco-based AI startup, which has confidentially filed for a US initial public offering, had previously warned about the hacking capabilities of its Mythos model and held it back from wide release.
On June 9, Anthropic rolled out a public version, called Fable 5, which included what it described as cybersecurity safeguards.
Anthropic worked with the government to test Fable 5 before it was released, the person close to the company said, and received its approval to deploy it.
US uses export control powers against Anthropic for first time
The letter to Anthropic said the Commerce Department was taking action through authorities granted it under the 2018 Export Control Reform Act to impose controls on emerging technologies essential to US national security. It marks the first time the Commerce Department has used that power, according to an export control expert.
The letter said that the Commerce Department would require a license for the export (or transfer to a foreign national in the US), and threatened that failure to comply with the new restriction would result in "prompt criminal and civil penalties."
However, export control experts said that AI models are generally not exported. They are deployed through remote access, which the export control regulations do not control, raising questions over whether Commerce has the legal authority to take such action.
Check if your data was exposed
Find out if your email and related personal information have appeared in known data breaches.
The Commerce Department did not respond to questions about the authorities in question. Neither the department nor Anthropic responded to requests for comment on the Monday meeting.
More than 80 cybersecurity executives and experts on Sunday signed an open letter to Lutnick and National Cyber Director Sean Cairncross that supported Anthropic’s position.
In that letter, cybersecurity leaders at major firms, including Nvidia and Adobe, asked the Trump administration to lift the restrictions on Anthropic.
FAQ
Did Anthropic Mythos escape?
No, there is no credible evidence that Anthropic’s “Mythos” model “escaped” or acted independently outside controlled systems.In the reporting you referenced, Mythos is described as a model that raised security and misuse concerns, leading to government restrictions and suspension of access — not something that broke out or “escaped.”So this is about access control and potential misuse risk, not an AI running free or escaping containment.
Will Anthropic Mythos ever be released?
Anthropic Mythos has been released, but not in a normal “open public rollout” way.Based on recent reporting, Mythos 5 exists but is restricted: it was released only to vetted partners and select users, while a safer derivative called Fable 5 was made broadly available instead.
Is Anthropic Mythos really that powerful?
Anthropic Mythos is extremely powerful.It is described as a frontier-level AI model that can autonomously find software vulnerabilities, generate exploits, and perform advanced cybersecurity and scientific reasoning tasks with minimal human input.
FAQ by nexos.ai, reviewed by Cybernews staff.
Unlock more exclusive Cybernews content on YouTube.