Trump export controls slam Anthropic’s new models

Trump export – An order from the Trump Administration pushed Anthropic to block foreign access to Fable 5 and Mythos 5, cutting off access entirely and triggering sharp, split reactions across the tech world. Critics called it baffling, incoherent, and disruptive for innovat

Anthropic’s announcement landed like a sudden door closing in a crowded lab: the Trump Administration ordered the company to block foreign access to its new AI models, Fable 5 and Mythos 5. The company responded by completely cutting off access to those models.

The move triggered immediate shock waves across the tech community, with reactions ranging from disbelief to alarm over how fast rules can change—and what those changes do to people building AI for a living.

Dean W. Ball. a senior fellow at the Foundation for American Innovation. said the situation. if true. is “baffling.” In his view. an administration that has a posture of exporting advanced AI chips to China while also pushing to ban “Britain (and every other non-American on Earth)” from using U.S. models doesn’t add up. He added: “I have no words.”.

Ball also framed the issue as a possible blend of strategy and law—“lawfare against Anthropic in particular” versus “extreme national-security hawkery”—then dismissed it as “simply cartoonish.”

Peter Girnus, a senior threat researcher at Zero Day Initiative, argued that the dispute can contain two truths at once. First. he pointed to how Anthropic “spent months marketing Mythos as too dangerous to release. ” and referenced Sam Altman’s statement that it was “incredible marketing to say we have built a bomb.” Girnus said the Commerce Department has now “formally agreed it is a bomb. ” adding: “If you describe your product as a munition in every press release. eventually a government takes you at your word.”.

Second, Girnus said the industry has run a similar experiment before. He recalled that in the 90s. the government classified encryption as a munition under ITAR. and that activists defeated those controls by publishing PGP’s source code as a book—arguing books are protected speech while floppy disks were arms exports. He added that even “a t-shirt with three lines of RSA Perl” was treated as a munition. and concluded that the controls collapsed because “math does not stop at customs.”.

Marc Andreesen, a partner at Andreessen Horowitz, zoomed in on a specific mechanic behind the cutoffs: the “deemed export” rule. Under that framework, showing controlled technology to a foreign national inside the U.S. counts as exporting it abroad. “Which is why Anthropic’s own foreign-national employees are now locked out of the model they built,” he said. “The munition is in the building and the people who made it are not allowed to look at it.”.

Chris McGuire. a senior fellow for China and emerging technologies at the Council on Foreign Relations. took a different starting point. He said targeted export controls on model access can be prudent. But he described the approach as “across the board controls on all countries on a single model. without any warning” and called it “highly questionable.” He added that imposing “equally broad deemed export controls” that also restrict access to foreign nationals is “just absurd—and obviously will result in the model being pulled from distribution. as just happened.”.

McGuire argued that export controls are a “critical tool” that can extend the U.S. lead in AI when used correctly. But he said that when used incorrectly. they “will stifle AI development.” He then criticized what he called incoherence in the Department of Commerce’s export control strategy: he said it is sending powerful AI chips to China. not enforcing controls to prevent Chinese smuggling. creating “massive loopholes” for AI chips to be sent to China.

and preventing U.S. AI companies from releasing their own models. His conclusion was direct: “This has to stop.” He said the U.S. needs a “smart export control strategy” that robustly denies adversaries access to advanced technology while “advantaging US companies.” He added that if the Bureau of Industry and Security (BIS) “doesn’t understand how to use its authorities or what the implications are” it needs new personnel to execute a competent strategy. calling the current

effort “incoherent and self-defeating.”.

Matthew Pines, CEO of Physical Superintelligence, warned that the situation could ripple far beyond Anthropic. He said it “this is gonna send shockwaves through every lab and neolab. ” and pointed to what he described as a “strict liability standard” in U.S. export control laws, calling it “a very sharp blade.”.

Dan Shipper, CEO of media and AI software company Every, suggested the immediate crisis might not last. “My take on this situation currently is that they’ll unban it in a few days and the net effect will be increased demand for Fable. ” he said. But he also emphasized the human cost of disruption inside companies. He called this kind of episode “extremely disruptive and distracting. ” and said the only comparable scenario he remembered was Sam Altman’s firing. which he said was resolved relatively quickly. Even then, Shipper said, it disrupted momentum for a while. He ended with a simple wish: “hoping for a good outcome here!”.

Josh Pigford, founder of Baremetrics, said Anthropic “has not done theirselves ANY favors” with what he described as “hyperbole” over the past 6-12 months. He also stated: “But I also guarantee this has zero to do with national security.”

Peter Barnett, a researcher at Machine Intelligence Research Institute, offered a warning about a possible escalation cycle. If the U.S. government controlled Fable because another company said it could jailbreak it. Barnett argued. the industry could be drifting into “a regime where AI companies red team each others’ models before the USG allows deployment.”.

Ryan Brewer, a member of technical staff at OpenAI, raised his own concern about concentration. He said that if the U.S. government continues on this path, “Eventually you will only be able to access frontier intelligence in a small set of buildings in the Bay Area.” “Shame,” he added.

Ketan Ramakrishnan, a Yale Law professor, framed the issue in broader political terms. He said the federal government is going to regulate AI developers aggressively. and the question is whether regulation will be intelligent—whether Congress and public deliberation will play a role. or whether it will be handled through “opaque executive action.”.

The immediate decision—blocking foreign access to Fable 5 and Mythos 5 and then cutting off access entirely—has left the tech world reacting not just to a policy, but to the shock of how quickly rules can reach into products, companies, and day-to-day work.

Anthropic Fable 5 Mythos 5 Trump administration export controls deemed export rule AI regulation BIS Commerce Department U.S.-China AI tech reactions foundation for American innovation Andreessen Horowitz Council on Foreign Relations Physical Superintelligence OpenAI Yale Law professor