ARTICLE AD BOX
The Trump administration just handed down a big directive to AI company Anthropic: block all foreign nationals from using its cutting-edge models, Fable 5 and Mythos 5. The official reason? National security. But there’s a catch—Anthropic says it can’t separate out users based on nationality without turning the models off for everyone. So that’s exactly what happened. The models are now offline across the board.
Commerce Secretary Howard Lutnick sent a letter to Anthropic’s CEO, Dario Amodei, laying out the terms. Now, any access, whether abroad or by non-citizens here in the U.S., is off-limits unless Anthropic gets an export license from the government. And those aren’t blanket permissions—each use needs its own individual approval. Messing up on this means serious penalties.
This shutdown happened just days after Anthropic debuted Fable 5 on June 9, the first public Mythos-class model. Mythos 5 itself—essentially the same model, just with less strict security—was already limited to trusted partner organizations working with Anthropic’s Project Glasswing. It’s not the first time things have gotten tense between the government and Anthropic.
Back in February, President Trump told all federal agencies to stop using Anthropic’s tech immediately. The Defense Secretary called them a national security risk. A judge put those punishments on hold in March, and the White House tried to work out some kind of deal to get government access to Mythos. In May, there was another fight—this time, the administration pushed back against Anthropic’s plan to roll out Mythos to a broader field, about 120 organizations.
All this time, there were leaks: reports about the NSA using Mythos for offensive cyber work, with Anthropic engineers on-site in secure facilities. When the latest export control order dropped, Anthropic said it was all a misunderstanding. They warned that if the U.S. stuck with this approach, no one would ever be able to launch new, advanced models in the country—pretty much freezing progress for everyone in the field.
In the meantime, they shut off both Fable 5 and Mythos 5 for all users, but left other products like Claude Opus 4.8 running. An administration official pointed to a security report from another company—one that claimed it broke through Mythos’s safety barriers—as the trigger for all this. The government tried to get Anthropic to stop these releases quietly. That didn’t happen, so they sent the formal order.
From the administration’s point of view, Fable 5 and Mythos 5 need to stay locked up until the national security framework gets tightened up. That process should wrap up in a few weeks—or so they claim—which is much longer than Anthropic was planning to keep Fable 5 free for subscribers. Their paid version was supposed to come out on June 22, but that’s out the window for now.
All this is just the latest sign that Washington wants to start treating top-tier AI like a key national security resource. Meanwhile, official rules are still pretty loose—a pre-deployment review is voluntary, and there’s no set licensing process in place yet. Still, the pressure is mounting.
English (US) ·