Trump export order puts Anthropic’s flagship models on ice

Trump export – Anthropic says a US export control directive it received Friday would force it to suspend access to its Mythos 5 and Fable 5 models by “any foreign national,” including foreign national Anthropic employees. The company spent the weekend in urgent talks with to

On a weekend when much of the country was celebrating the USA’s first World Cup win and the New York Knicks championship, Anthropic was focused on something harsher: a federal order that could freeze two of its newest AI models just as the company was pushing them out.

At 5:21 PM on Friday. Anthropic received a US export control directive telling it to suspend access to its Mythos 5 and Fable 5 AI models by “any foreign national” inside or outside the US. “including foreign national Anthropic employees.” Anthropic concluded that there was only one way to comply—by completely disabling products it had spent the prior week hyping. The scramble that followed wasn’t just technical. It meant travel to Washington, DC and direct attempts to change President Donald Trump’s mind.

Anthropic also described Mythos 5 and Fable 5 as built on the same foundation as Mythos Preview. which Anthropic had dubbed too dangerous to publicly release. Mythos 5 was made available to a select group of government agencies and companies. while Fable 5—described as featuring additional safeguards—was deemed “safe for general use.” The sudden pivot came after a report suggested those guardrails may have failed. and Anthropic’s own earlier warnings about Mythos falling into the wrong hands returned with force.

A source familiar with the situation. who participated in negotiations between Anthropic and the Trump administration. said the administration called the AI lab on Friday around 1pm ET and gave a 90-minute ultimatum: shut down access to Mythos 5 and Fable 5. If Anthropic didn’t, the government would impose export controls on the company by authority of the US Commerce Department.

Within minutes, the response moved straight to the highest level. The source said Anthropic executives were talking to the White House within 15 minutes of that first call. and that CEO Dario Amodei joined the discussions about an hour and 15 minutes after the initial call. Amodei directly spoke with US Treasury Secretary Scott Bessent. Commerce Secretary Howard Lutnick. and National Cyber Director Sean Cairncross. in some cases more than once. the source confirmed.

In its own Friday release. Anthropic said it believed the government “believes it has become aware of a method of bypassing. or ‘jailbreaking’ Fable 5.” But Anthropic pushed back on how that threat was being framed. The company said the jailbreak in question was a “potential narrow. non-universal” one that was “shared with the government” by an entity the company declined to name. Anthropic also said the behavior wasn’t unique to Fable 5. writing that it had “reviewed a report” it believed was the basis of the directive and validated that the level of capability shown there is widely available from other models. including OpenAI’s GPT-5.5.

While the company was presenting its case, the administration’s concerns were being placed under a different kind of spotlight.

Semafor reported. citing one source familiar. that the hubbub began because the US government was concerned that a China-linked group had accessed the technology. But the source familiar with the negotiations said the China rumors went back weeks and referred to a large global telecommunications company that had initially been cleared for inclusion in access to Mythos Preview. The source said that when the US government shared its concerns, Anthropic immediately revoked access.

In the public conversation. David Sacks— the US government’s former AI and crypto czar who stepped down in March—posted on X. His post did not mention China. Sacks did. however. write about the unnamed entity that exposed the issue to the government. describing it as “a highly credible trusted partner of both Anthropic and the USG who was testing Fable [which] came forward with a jailbreak of those guardrails.”.

Other threads complicated the picture. Some reports pointed to Amazon CEO Andy Jassy as the person who flagged concerns to the US government after researchers at Amazon had red-teamed Fable 5. That conclusion conflicted with some independent red-teamers who said they were impressed with the level of protections. Still. the source familiar with the negotiations said Amazon’s research was explicitly mentioned in conversations with the US government. and that Anthropic had had access to the paper within days of the Friday export control directive. The source said Anthropic then went back-and-forth with Amazon researchers to discuss it. adding that everything in that paper could be achieved by OpenAI’s GPT-5.5.

With the clock running on government access restrictions, Anthropic spent the weekend scrambling to repair the relationship. The company began with virtual meetings and then flew employees to DC. including Dave Orr. Anthropic’s head of safeguards; Logan Graham. who runs its frontier red team and has led work on Project Glasswing; and Nicholas Carlini. a leading frontier developer and cybersecurity researcher.

Axios. citing a source familiar with the Trump administration’s thinking. reported that the company had repeatedly made missteps in communicating with the administration and “has not done a great job at trying to speak to the administration and appreciate the ideological differences.” For Anthropic. the timing was especially punishing: the company had banked on Mythos to help it recover from months of high-profile clashes with the US Department of Defense.

The source familiar with the negotiations said Anthropic pre-briefed the administration on Fable 5. and that the US Department of Commerce conducted testing pre-deployment with no concerns shared at the time. The source also said Anthropic had been working closely with government agencies since Mythos Preview’s release.

There’s a political story underneath the policy language now, and it’s one that’s drawn criticism from unexpected corners of the tech world.

The Trump administration. the reporting says. initially took a hands-off approach to AI safety—but after Mythos. it has become more ambivalent. It has continued to fret about the threat of losing the AI race to China. even as it frets in the direction of stricter control measures. The decision has energized international calls for alternatives to American AI systems and. in the immediate term. effectively put a major US AI company’s new flagship model on ice.

By Sunday, the pressure turned into a public push.

A public letter from tech and cybersecurity executives called for restrictions on Fable 5 to be repealed. The letter states: “Not all of us agree that AI regulation is the right way forward. ” and adds that if regulations are going to happen regardless. then they should be rooted in “scientific evaluations developed with input from industry and academia.”.

Alex Stamos. chief product officer at Corridor. told The Verge he organized the public letter because the countless number of vulnerabilities in the past decade-plus—written in a variety of different coding languages—mean AI needs to patch before bad actors find them. Stamos said. “We’re in a race. and I think policymakers don’t understand that. ” adding that there’s “this weird arrogance” in the idea that American labs are hugely ahead of adversaries and will always be. He argued the gap is measured in months and said, “And you can see that in the open evaluations. The cutting-edge models are only something like six months ahead of the Chinese models — and those are the models we know about.”.

The letter also argues that although Anthropic’s Mythos-class models are skilled at finding cybersecurity vulnerabilities and taking advantage of exploits. they aren’t “uniquely good” at these tasks. It says Fable 5’s safeguards were “so aggressive as to be the source of humor in the cyber community on launch day.” Stamos told The Verge that “there’s a real overstatement of Mythos’ capabilities” and that Anthropic is “somewhat responsible for this themselves.” He said. “Mythos is great. but the real turning point was really last year.”.

Stamos said the industry is already adjusting in response to political risk. He pointed to backup contracts being signed with non-US companies and open-weight models being deployed on alternative hardware arrangements. arguing that the weekend made political risk a factor in company planning more than ever before.

Ben Van Roo. co-founder and CEO of Legion Intelligence— a system of agents for the national security community—offered a different kind of blunt reaction. Van Roo told The Verge that the directive of “no foreign national should use this model” is the most impossible thing to enforce. He said his AI community network was “exploding” when he first read it.

The stakes widen further when the policy intersects with competition.

OpenAI, Google, and Microsoft have all released their own comparable products to Anthropic’s Mythos. The concern. in this telling. is that if the Trump administration bans Anthropic’s advanced cybersecurity models. it can make a case for banning competitors’ models too. That could push AI companies toward a choice: unite to help Anthropic. or present themselves as safer and more compliant alternatives—especially in a political climate that once treated autonomous weapons as something that could spark high-stakes fights.

Even as the administration tries to free tech companies from regulatory hassles. the Anthropic order could become a dramatic restriction on powerful AI models depending on what comes next. Van Roo called it “uncharted territory” in the regulatory setting. and said he doesn’t think it will be the last time something like this happens.

There is also a broader cultural pressure described in the reporting: the rise of “AI populism. ” with people pushing back against the industry’s influence through protests at data centers. pledges to quit using AI chatbots. lawsuits over wrongful deaths. and even attempted attacks on AI company CEOs. Van Roo said the Trump administration’s moves against Anthropic could stoke “greater fears and concerns. potentially for the wrong reasons.”.

For now, the talks are described as constructive—yet unresolved. The source familiar with the negotiations said some members of the administration admitted that putting export controls on model providers isn’t ideal because competitors with similar products may end up under the same restrictions. The source also said the US government is currently exploring a program that would encourage the export of American AI systems.

Monday’s talks concluded with no resolution as of yet.

And while Anthropic continues negotiating with the US government, other conflicts loom. There is “little chance” Anthropic’s other issues with the US Department of Defense won’t resurface—particularly the ongoing battle over acceptable usage policies for Anthropic’s tech by the US military.

Van Roo said the situation is new. with “anything potentially this drastic” being unprecedented. and that it carries real ramifications for enforcement. He asked a question that cuts to the center of the dispute: “Who gets to use this new technology that continues to outpace our own ability to regulate it?”.

Anthropic Mythos 5 Fable 5 US export control directive AI regulation Dario Amodei Scott Bessent Howard Lutnick Sean Cairncross Mythos Preview cybersecurity safeguards jailbreak GPT-5.5 Legion Intelligence Alex Stamos