Anthropic executives met with administration officials over the weekend after the government directed generative AI vendor to suspend access to some of its newest AI models.
A June 12 statement on Anthropic’s website revealed that the Trump administration, citing national security authorities, had issued an export control directive to suspend access to Fable 5 and Mythos 5 by any foreign national, inside or outside the U.S., including foreign Anthropic employees.
As a result, Anthropic said the only option to ensure compliance was to deny access to all customers.
The statement said the government had not provided specific details about its concern. However, it said: “Our understanding is that the government believes it has become aware of a method of bypassing, or ’jailbreaking’ Fable 5.”
This jailbreak would override a built-in safeguard, potentially opening Fable 5 up to misuse by bad actors. This has been an ongoing concern about Mythos, the underlying AI model on which Fable 5 is built. Anthropic restricted access to Mythos because of fears over its powerful ability to identify flaws in major operating systems and web browsers.
While Anthropic has complied with the government’s directive, it has made clear that it does not agree with the rationale behind it, claiming it has only been provided “verbal evidence of a potential narrow, non-universal jailbreak.” This consisted of asking the model to read a specific codebase and fix any flaws.
The vendor pointed out that no testers have yet been able to find a universal jailbreak, which would potentially have much greater cybersecurity repercussions, and said: “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model developments for all frontier model developers.”
Speculation over how and why the government has acted has gained momentum since Anthropic, which is currently pursuing an IPO, made its statement. Earlier this year the Trump administration had a public fallout with the company, labeling it a supply chain risk to national security.
Rumors are rife, too, over how the government was made aware of the potential for jailbreaking, with some outlets reporting that Amazon had raised concerns.
The response to the move has been mixed, with some industry observers bemused. Dean Ball, an AI policy expert who once served in the Trump administration, posted on X, formerly Twitter: “I can’t tell if this is lawfare against Anthropic in particular or extreme national security hawkery. Regardless, it is simply cartoonish.”
Meanwhile, in Europe, some political leaders voiced alarm that AI tech their countries are becoming increasingly reliant on could simply be “switched off” at the behest of Trump.
Anthropic’s statement added that no other models were affected, and it was working to restore access to Fable 5 and Mythos 5 soon.
