Trump Says Anthropic Is No Longer a National Security Threat, One Week After Ordering Its AI Models Shut Down

Since the Trump administration issued a shutdown directive against Anthropic's newest AI models on June 12, the situation has shifted significantly: Trump told Axios on June 19 that Amodei and his company have behaved "very responsibly" and no longer pose a threat.

What Trump Said

In an interview with Axios's Marc Caputo that aired June 19, Trump was asked directly whether he still viewed Anthropic and CEO Dario Amodei as a national security threat. His answer: "Well, not now, but a week ago, maybe."

He credited a meeting with Amodei at the G7 summit this week for his change of view. "He responded to us very quickly, because you know it's tremendous liability. You can't play games with it. And he responded very responsibly," Trump said, according to The Epoch Times via ZeroHedge.

What Triggered the Crisis

On June 12, U.S. officials directed Anthropic to cut off foreign nationals' access to its two newest models: Claude Fable 5 and Mythos 5. Anthropic, rather than selectively restricting access, suspended those models for all users entirely.

The company said it received no specific details about the alleged threat. Its public statement at the time: "Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5."

Fable 5 had launched to the general public on June 9. At launch, Anthropic described it as exceeding "the capabilities of any model we've ever made generally available," citing "exceptional performance in software engineering, knowledge work, vision, scientific research" and other domains.

A model that clears those benchmarks is also a model with serious dual-use potential. The jailbreaking concern, whatever its specifics, was not implausible on its face.

A Competitor Pulled the Alarm

Trump added one detail that hasn't gotten enough attention: he told Axios that one of Anthropic's competitors was the source of the initial complaint. "They didn't like what [Anthropic was] doing. They were very concerned," he said.

He did NOT name the competitor. This raises an obvious question: was the national security concern genuine, or was a rival AI company using regulatory channels to kneecap a competitor's product launch? Trump's phrasing doesn't settle that question—it opens it.

Anthropic did not respond to a request for comment from The Epoch Times before publication.

The Strongest Counterargument

Critics of how the administration handled this have a legitimate point. The government issued a shutdown directive with no specific articulation of the threat, forced a private company to cut off its own customers, and then after a summit meeting with the CEO declared everything fine. No public explanation of what the jailbreak method was. No public explanation of what Anthropic did to fix it. No independent verification that the threat was real rather than competitor-driven.

If the national security risk was genuine enough to justify a compulsory shutdown on June 12, the public deserves more than a presidential interview quote explaining that everything is okay now. That's not how security disclosures are supposed to work, and the opacity here is a legitimate governance concern regardless of which party is in power.

What's Verifiable and What Isn't

What is confirmed: the June 12 directive happened, Anthropic suspended access, and Trump met with Amodei at the G7. What is NOT confirmed: the specific nature of the jailbreak vulnerability, whether it was ever patched, what Anthropic agreed to do differently, and who the competitor was that raised the alarm.

The administration has not announced any formal investigation, and no charges or legal proceedings have been filed against Anthropic or any individual. This was an informal directive, not a court order, which means there is no public legal record to examine.

Where This Leaves the AI Industry

The episode sets an uncomfortable precedent. A competitor apparently had enough access to administration officials to trigger a product shutdown—one that cost Anthropic users and, presumably, revenue—without any public process. The shutdown lasted roughly a week before a G7 handshake effectively resolved it.

The unresolved question as of June 20: Fable 5 and Mythos 5 access was suspended for all users as of June 12. Anthropic has not publicly announced whether those models are back online, under what conditions, or what technical changes, if any, were made before access resumes.