Cybersecurity Policy Report, Anthropic, White House Aim to Resolve Dispute Over Foreign Access to AI Cyber Capabilities, (Jun 15, 2026)
Anthropic and the Trump administration are trying to resolve a dispute that led Anthropic to end public access to two of its most advanced AI models over the weekend.
Late Friday, Anthropic revealed that the Commerce Department ordered it to prevent foreigners from using its Fable 5 and Mythos 5 models. Since the company was unable to determine which users in the U.S. and abroad should lose access to the models, it was forced to “abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” it said in a statement.
The Commerce Department and White House didn’t specify the reasons for the directive to Anthropic, but David Sacks, co-chair of the President's Council of Advisors on Science and Technology indicated in a social media post that the administration was concerned that the Anthropic models would provide “advanced cyber capabilities to people who shouldn’t have them.”
The Fable 5 and Mythos 5 models are based on Anthropic’s Claude Mythos2 Preview, which proved to be so adept at finding and exploiting cyber vulnerabilities that Anthropic withheld it from public release. Those hacking capabilities were stripped from the publicly released Fable 5 and Mythos 5, Anthropic said.
Based on “a number of conversations with folks inside and outside government about the current situation with Anthropic,” Mr. Sacks said the guardrails Anthropic added to the public models were being circumvented.
“A highly credible trusted partner of both Anthropic and the USG [U.S. government] who was testing Fable came forward with a jailbreak of those guardrails,” he said, and when the administration asked Anthropic Chief Executive Officer Dario Amodei “to fix the jailbreak or de-deploy the model,” Mr. Amodei “refused.”
The administration’s “hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority,” Mr. Sacks said.
Anthropic’s description of events, however, contradicts Mr. Sacks’ narrative. The vulnerabilities in Fable 5 that appear to have prompted the Commerce Department directive “all appear relatively simple” and are present in public models offered by other AI developers, it said.
“We have instituted strong safeguards that greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity (among others). In fact, our safeguards are so strong that many users have complained that they are overly broad,” the company said.
Before Fable’s public release, Anthropic consulted the U.S. government, the United Kingdom’s AI Security Institute, “multiple private third-party organizations and internal teams to red-team Fable’s safeguards for thousands of hours in total,” the company said. “These tests showed that Fable’s safeguards are substantially more effective than those of any previously deployed model.”
Anthropic representatives were reportedly in talks today with administration officials to identify and fix any concerns that led to the Commerce Department directive. Anthropic is viewed as one of the more responsible AI developers, and its scruples regarding how its models are used previously led the Trump administration to try to cancel its government contracts and designate it as a national security threat (CPR, March 19).
The hacking capabilities of frontier AI models from Anthropic and OpenAI recently prompted the White House to reverse its mostly hands-off policy toward AI development and release an executive order that encouraged developers to submit for the government’s review any new models that might pose national security threats (CPR, June 2).
Rep. Ted Lieu (D., Calif.) criticized the administration’s approach to AI in a social media post over the weekend. Administration officials “have no idea what they are doing on AI,” he said. “They first let the AI companies run wild. Then they did a weak EO that is voluntary. Then today they went Oh Shit and took an action that caused Anthropic to take Mythos 5 and Fable 5 offline.”
Alan Rozenshtein, a law professor at the University of Minnesota Law School, described the Commerce Department’s use of its export-control authorities to block access to an AI model as a “kill switch” for frontier AI.
“Although the legal basis for the order has not been made public, Commerce is most likely relying on the Export Administration Regulations (EAR), which its Bureau of Industry and Security (BIS) administers under the Export Control Reform Act of 2018,” Mr. Rozenshtein noted today in a post on the Lawfare Institute blog.
“To avoid the delay of notice-and-comment rulemaking, BIS can send a company an ‘is informed’ letter that imposes a license requirement immediately, with no notice-and-comment rulemaking,” he explained.
“In the short term, such a control might counterproductively bar model access to some of the very researchers needed to fix the problems the government is worried about. In the long term, it would make it very difficult for Anthropic and other frontier AI companies to secure the foreign talent that is key to AI's development, thus threatening America's position as the leading AI power,” he warned.
“If Washington is going to treat frontier models as national-security assets it can switch on and off, Congress will have to step in and establish a proper framework with meaningful standards and a defined process,” Mr. Rozenshtein added.
MainStory: TopStory FederalLegislation LitigationEnforcement DataSecurity AINews