Concerns from the White House regarding potential access to the Mythos model by a China-linked group were among the reasons for export restrictions on Anthropic. This was reported by Semafor, citing a source.
At the same time, Business Insider and Axios identified another trigger as reports of bypassing protections on Fable 5. Anthropic itself claims that the method in question was "narrow" and not universal, lacking unique capabilities for the model.
What Triggered the Action
According to media reports, the Trump administration was alarmed by suspicions of access to Mythos by a China-linked group. Journalists did not specify which organization was involved, how it occurred, or how the White House learned of the incident.
According to The Verge, the White House has not publicly confirmed this version. An Anthropic representative told Semafor that the topic of "Chinese access" to Mythos was not raised in discussions with authorities. The company prohibits access to its products from China.
In Business Insider, it was noted that on June 11, Amazon CEO Andy Jassy informed the presidential administration about the possibility of bypassing Fable 5 restrictions. According to a source from the publication, the corporation was responding to a request to verify the model. According to Axios, U.S. Secretary of Commerce Howard Latnik sent a letter to Anthropic CEO Dario Amodei stating that Fable 5 and Mythos 5 were subject to export controls outside the U.S. and for all foreign individuals within the country.
Officials allegedly tried to persuade Anthropic to voluntarily suspend access to the models, but the company refused. A source close to the startup disputed this version, claiming that the White House gave the startup 90 minutes to disable the models without detailing the nature of the threat.
On June 13, White House advisor David Sacks wrote on X that the administration received a warning about a potential bypass of Fable 5 protections from a partner trusted by both the authorities and Anthropic.
"The administration hopes that Anthropic will resolve the security issue, the export control will be lifted, and Fable will return to general access. [...] The ball is in Anthropic's court," Sacks wrote.
Anthropic's Position
Anthropic stated that the government letter did not contain specific details about national security risks. According to the company, the concerns were related to the method of bypassing Fable 5 protections.
The startup claims to have examined a demonstration of this technique. In its assessment, it allowed for the identification of "a small number of previously known minor vulnerabilities." Anthropic added that other publicly available models can identify the same issues without bypassing protections.
The company also stated that before launching Fable 5, it spent "thousands of hours" testing protections alongside the U.S. government, the UK’s AI Security Institute, third-party organizations, and internal teams. According to Anthropic, testers have not yet found a universal method for mass bypassing of the model's restrictions.
"We consider this a misunderstanding and are working to restore access as soon as possible," the company stated.
On June 9, Anthropic introduced Fable 5 and Mythos 5. The company described Fable 5 as a Mythos-class model for general access. Mythos 5 uses the same base model but has relaxed restrictions in certain areas and is intended for cybersecurity professionals and critical infrastructure providers as part of Project Glasswing.
On June 2, Anthropic announced the expansion of Project Glasswing to approximately 150 organizations across more than 15 countries. According to the company, early partners in the program had identified over 10,000 high and critical severity vulnerabilities by that time.
It is worth noting that following the U.S. export directive requiring Anthropic to limit access to Fable 5 and Mythos 5 for foreign nationals, tokens of decentralized AI projects Venice and Morpheus rose by 14% and 21%, respectively.