The Trump administration is reigniting its feud with the latest AI models
The dispute between Anthropic and the Trump administration escalated again on Saturday after the administration took the unusual step a day earlier of demanding that the artificial intelligence company cut off foreign access to its latest models, as top officials indicated the dispute was unlikely to be resolved quickly.
Late Friday, Anthropic revealed in a blog post that it had banned all customers from accessing its most advanced artificial intelligence systems, known as Fable 5 and Mythos 5, after receiving an order from the administration to suspend access to all foreign nationals. The move shocked former US officials and cybersecurity experts, many of whom questioned the validity of the action, noting that it departed from the casual approach to policing the booming AI industry that President Trump endorsed earlier this month.
Antropic said the directive does not explain the national security concerns that prompted it. However, the company added that the government said it had learned of a “jailbreak” method, or bypass of security restrictions, on the Fable 5 that was intended to limit a customer’s ability to misuse the product for hacking or other potential harm. Anthropic countered that concerns about the jailbreak method were overblown.
The announcement stoked tensions between the Trump administration and Anthropic, which earlier this month filed confidentially for an initial public offering following a funding round that valued it at nearly $1 trillion. The two sides have argued for months over how Anthropic’s AI systems could be used in the military and intelligence services, culminating in the Pentagon labeling the company a supply chain risk.
On Saturday, top Trump administration officials and allies of Mr. Trump stepped up their attacks on Anthropic.
“Three months ago, @DeptofWar kicked @AnthropicAI out of our building – forever,” Defense Secretary Pete Hegseth said on X. “Each day proves why it was the right move.”
David Sacks, a venture capitalist who until recently worked in the administration as the AI czar, accused Anthropic of recklessness in releasing its latest model, called Fable 5, in a lengthy social media post on Saturday.
Mr. Sacks, who said he had spoken to many people inside and outside the administration about the directive, said the administration asked Dario Amodei, Anthropic’s chief executive, to address the jailbreak issue after a “highly trusted trusted partner” of Anthropic and the government came forward with the research.
“Dario refused,” Mr. Sacks said.
A person briefed on Mr. Amodei’s conversations disputed that account, saying Anthropic was happy to address concerns.
Mr. Sacks did not name the “trusted partner” he referred to in his post. However, several tech firms, including Amazon, have spoken to the White House about security issues, according to people familiar with the matter. Several people said the message from Amazon CEO Andy Jassy detailing the security issues was the most influential, raising concerns about the capabilities of the new Anthropic model.
But several of those officials said a separate document from Amazon explaining safety concerns with the Anthropic model was misleading. The capabilities that the document highlights in the Anthropic model are also present in OpenAI’s top model, 5.5.
An Amazon spokesman declined to comment on the exact nature of its discussions with the White House. “It is not unusual for governments to seek our advice on potential security risks,” the spokesman said. “We do not share the details of those discussions when they occur.”
Administration officials called Antropia officials at 1:15 p.m. Friday and gave them 90 minutes to download their most advanced models, citing an undefined national security concern, according to people briefed on the discussions. Anthropic officials asked for more information, trying to figure out exactly what the issue was, as the Department of Commerce’s review and testing of Fable revealed no significant concerns.
Then, at 5:21 p.m., Anthropic was notified that the Trump administration was implementing export controls that effectively forced it to withdraw its model, which consumers had just begun to access.
Updated
Discussions to resolve the dispute continue. Howard Lutnick, the commerce secretary, spoke with Antropia officials on Friday and was scheduled to have another meeting with senior company officials on Saturday night, according to people briefed on the discussions and plans.
Some experts said the Trump administration either misunderstood or deliberately misunderstood what happened. Katie Moussouris, CEO of Luta Security, said on social media that she saw the research paper that prompted the administration’s action.
“This is not a jailbreak,” she said, but rather a defensive maneuver designed to limit misuse of the model. “If national defense is the goal, that’s an own goal.”
Earlier this month, the Trump administration issued an executive order asking tech companies to voluntarily let the government review their new models before releasing them to the public. But it did not give the government an official role in approving their release.
The order was delayed after a fierce debate in which AI companies — and some sympathetic government officials — pushed back against government interference in rolling out the model. The dispute came on the heels of the very limited release of Anthropic’s Mythos, an artificial intelligence model so good at detecting and weaponizing new cybersecurity vulnerabilities that it sparked widespread concern about its potential to wreak havoc.
The new restrictions on Anthropic have raised questions about the implications for other AI models that may have similar capabilities. Former officials and technology experts said the administration did not seem to think through the long-term implications of such a move.
But the measure could be limited to Anthropic, which has received a series of attacks from the administration in recent months.
In February, amid the Pentagon’s clash with Anthropic, Mr. Trump the company called the “radical left, woke society” and the “leftist lunatics” who seek to dictate how the government wins and fights wars.
“I am directing ALL Federal agencies in the United States Government to IMMEDIATELY Cease all use of Antropic technology,” the president wrote on social media. “We don’t need it, we don’t want it, and we won’t do business with them again!”
Some administration officials have been looking for a way out of the dispute in recent weeks, US officials say. White House and intelligence officials have pushed through a classified contract between Antropic and the National Security Agency that would allow the spy agency to use the company’s technology for a variety of purposes, including intelligence analysis and uncovering new computer vulnerabilities.
Officials at the NSA, which is responsible for digital wiretapping and government cybersecurity, were not involved in Friday’s decision, according to people familiar with the matter.
Many US officials have said that the new technology Anthropic has developed is too important to national security to allow a dispute with the Pentagon to block cooperation. But in the top ranks of the Pentagon, officials remain angry with the company, insisting that the firm sign the same contractual provisions that other AI companies have accepted.
In a June 12 letter seen by The New York Times, Mr. Lutnick told Mr. Amodei that the company would require a special license to distribute its Mythos and Fable 5 models “to all destinations around the world,” as well as to share them with non-U.S. citizens.
“Failure to comply will result in immediate criminal and civil penalties as provided by law,” Mr. Lutnick wrote.
The government has scrutinized AI models before, but past restrictions have been more targeted. The measure, introduced by the Biden administration early last year, limited companies’ ability to export so-called weights for specific AI models — proprietary numeric values that tell the model how much importance to place on various data.
The measure taken by the Trump administration goes beyond the fact that Anthropic cannot share the model with any foreign country or foreign national without first obtaining a license.
Limiting the involvement of foreign nationals, even those with green cards, could be particularly chilling in an industry that relies on foreign talent. In the past, government officials often reserved this type of restriction for the most sensitive technologies, such as weapons systems.
Chris McGuire, a former technology official in the Biden administration who is now a senior fellow at the Council on Foreign Relations, said the administration’s current policy is that AI chips can be exported to China, but that Canadian green card holders in the United States cannot access leading American models.
“That’s absurd,” he added.