Onpode
Cover art for Trump just reversed course — Anthropic is no longer a national security threat to his administration

Trump just reversed course — Anthropic is no longer a national security threat to his administration

June 22, 2026 · 6 min

Maya Chen & Dr. Nathan Hayes

Trump reversed his designation of Anthropic as a national security threat seven days after the Bureau of Industry and Security issued an undisclosed directive that forced Anthropic to shut down its Fable 5 and Mythos 5 models globally within 90 minutes — with no public legal basis given at designation or reversal.

On June 12, 2026, the U.S. Commerce Department issued an export control directive ordering Anthropic to restrict foreign nationals from accessing its two most advanced AI models, Fable 5 and Mythos 5, citing national security concerns. In response, Anthropic disabled access to both models for all users globally while it worked to comply.

0:005:36
Make your own on Onpode

Describe any topic. Hear it in minutes.

More Onpode episodes on anthropic׳s update

About this episode

On June 12, 2026, the U.S. Commerce Department issued an export control directive ordering Anthropic to restrict foreign nationals from accessing its two most advanced AI models, Fable 5 and Mythos 5, citing national security concerns. In response, Anthropic disabled access to both models for all users globally while it worked to comply.

Frequently asked

Why did Trump say Anthropic is no longer a national security threat?

Trump reversed his national security designation of Anthropic after meeting CEO Dario Amodei at the G7 summit in Évian-les-Bains, telling The Axios Show two days later that Anthropic 'behaved very responsibly.' No legal or factual basis for the original designation or its reversal was ever publicly disclosed.

What happened to Anthropic's AI models under the US export control directive?

The Bureau of Industry and Security issued an undisclosed directive on June 12th citing unspecified national security concerns. Within 90 minutes, Anthropic disabled Fable 5 and Mythos 5 for all customers globally — including its own employees — because its systems had no mechanism to filter users by nationality.

What triggered the US export control action against Anthropic?

The Bureau of Industry and Security's directive gave no specific details. Anthropic publicly stated it believes a guardrail bypass triggered the action, but the government never confirmed that. Anthropic's own explanation for why the shutdown occurred is, by its own account, a guess rather than confirmed fact.

What precedent does the Anthropic export control reversal set for AI companies?

The Anthropic case establishes that the expected response to an undisclosed government directive is immediate, total compliance — not a legal challenge or scope negotiation. Both Anthropic and the Trump administration publicly praised the full shutdown as the responsible standard, with no disclosed factual basis required at either end.

Can US export control laws effectively regulate AI models like they regulate hardware?

The Export Administration Regulations were built for discrete physical objects like semiconductors that customs agents can intercept at borders. Applying them to AI models like Fable 5 and Mythos 5 is structurally mismatched: controlling access requires constant technical enforcement that breaks down at scale, making the Cold War-era framework poorly suited to frontier AI products.

Grounded in 12 sources
Trump tells Axios he no longer views Anthropic as national security threat - Reuters · reuters.com
Trump tells Axios he no longer views Anthropic as a national security threat - CNBC · cnbc.com
Trump tells Axios he doesn’t see Anthropic as U.S. security threat - Los Angeles Times · latimes.com
Exclusive: Trump tells "The Axios Show" that Anthropic was a national security threat - Axios · axios.com
Anthropic suspends all access to Mythos model after US government bans foreign nationals use - CNN · cnn.com
White House’s Anthropic move jolts Congress back into the AI debate - Politico · politico.com
Inside the White House's AI power center - Yahoo · yahoo.com
White House Considers Vetting A.I. Models Before They Are Released · nytimes.com
US export ban on Anthropic's AI models further strains ... · aljazeera.com
Anthropic disables new AI model after White House security directive | PBS News · pbs.org
US curbs Anthropic AI access, raising global concerns · dw.com
Cybersecurity executives urge the Trump administration to ease restrictions on Anthropic AI models - AP News · apnews.com
Read transcript

Dr. Nathan Hayes: You know what I keep thinking about? The 90 minutes.

Maya Chen: That's — yeah, that's the number that stopped me too.

Dr. Nathan Hayes: Now, here's what we know: June 12th, Bureau of Industry and Security — that's the Commerce Department's export control arm — issues a directive to Anthropic. Unspecified national security concerns. Never disclosed publicly. And within 90 minutes, Anthropic disables Fable 5 and Mythos 5 for every customer, globally. Not just foreign nationals — everyone, including their own employees.

Maya Chen: Because they couldn't — wait, they literally couldn't filter by nationality, right? That was the reason?

Dr. Nathan Hayes: Correct. The compliance mechanism didn't exist. So the response was total shutdown.

Maya Chen: And then five days later Trump is at the G7 in Évian-les-Bains, he meets Dario Amodei, and two days after that he's on The Axios Show telling Marc Caputo Anthropic 'behaved very responsibly' and he basically doesn't see them as a threat anymore. 'Well, not now. But a week ago, maybe.' That's the whole arc.

Dr. Nathan Hayes: Seven days. Designation to reversal — seven days, no disclosed legal basis at either end.

Dr. Nathan Hayes: Now, here's what's actually new — and I mean structurally new, not just politically dramatic. The Export Administration Regulations, the framework the Bureau of Industry and Security enforces, were built for discrete physical objects. Semiconductors. Encryption hardware. Things customs agents can literally intercept at a border crossing.

Maya Chen: Like, a hard drive in a suitcase.

Dr. Nathan Hayes: Exactly. Now imagine asking those same customs agents to stop someone from *thinking about* what's on the hard drive. That's what applying Export Administration Regulations to Fable 5 and Mythos 5 looks like. The tool has no grip. A model has no discrete physical form — controlling access requires constant technical enforcement, and that breaks down at scale. This is the first high-profile application of that Cold War framework to a frontier lab's commercial products, and the mechanism is structurally mismatched to the object.

Maya Chen: So the actual headline isn't the reversal — it's that this tool was used at all.

Dr. Nathan Hayes: Right. And here's what TechCrunch and Zack Whittaker reported — enforcement wasn't about a jailbreak. An AI model guardrail bypass. Anthropic said publicly it *believes* that's what triggered it, but the government never confirmed that. The directive gave no specific details. So Anthropic's own explanation for why this happened is... a guess.

Maya Chen: Howard Lutnick runs the department that sent that letter and said nothing. Publicly. That's — yeah. That's the story.

Dr. Nathan Hayes: Now the take I want to name — the one circulating — is that Anthropic navigated this brilliantly. Dario Amodei flies to Évian-les-Bains, Trump finds him nice and smart at the G7, models come back online, everyone wins. That framing is wrong.

Maya Chen: Mm. But wait — the Defense Production Act was on the table. Trump literally told Marc Caputo he has 'the power to use a lot of things.' If that's the alternative, isn't swift de-escalation... actually rational?

Dr. Nathan Hayes: Rational for Anthropic, yes. Good governance? No. Think about the London research team — they wake up Tuesday, Fable 5 is simply gone. No explanation. No timeline. No recourse. And the resolution to that isn't a legal process — it's whether Trump found the CEO likable over lunch.

Maya Chen: That's — yeah. That's not a system.

Dr. Nathan Hayes: Anthropic's own statement called this a 'shared goal' with the administration. Trump praised them for responding 'very quickly' and 'very responsibly.' Both parties publicly endorsed complete capitulation as the gold standard. Responsible toward whom, exactly? The directive had no disclosed factual basis — Anthropic said it *believes* a guardrail bypass triggered it. That's not knowledge. So what they're being praised for is surrendering to something they couldn't even — I mean, they didn't actually know what they were accused of.

Maya Chen: And now every company watching this knows: the expected response to an undisclosed directive is complete, immediate surrender. Not a legal challenge. Not even scope negotiation. That precedent — I just can't move past it.

Maya Chen: And the government never said what threat was actually resolved. Like, was it resolved? Did something change between June 12th and June 19th, or did Trump just... meet someone he liked at Évian-les-Bains and decide the concern wasn't real anymore? Because those are completely different outcomes and we're being asked to treat them as the same.

Dr. Nathan Hayes: No disclosed legal basis on June 12th, no disclosed legal basis on June 19th. The designation appeared, it lifted. Ad hoc governance — imposed and reversed within seven days with no public explanation at either moment. That's not a resolved threat. That's an unresolved one that someone stopped paying attention to.

Maya Chen: So if the Bureau of Industry and Security sends that same letter tomorrow — to the next company, whoever's running the next Fable 5 — what does the Anthropic case actually tell them to do?

Trump just reversed course — Anthropic is no longer a national security threat to his administration · Onpode