Anthropic has taken two of its products offline in response to new US export controls that for the first time target software-level restrictions. The company pulled Fable 5 and Mythos 5 for all users, a move that signals how far regulators can reach into the code of AI tools.
A directive that targets code, not just hardware
The directive from US authorities breaks new ground. Instead of focusing on chips or hardware, it goes after the software itself. That makes it a precedent for how AI models and platforms can be controlled at the source. Anthropic’s decision to remove the two products entirely suggests the controls left no room for regional restrictions or geo-blocking workarounds.
Global access gets squeezed
For developers and users outside the US, the move tightens the screws. Export controls have historically targeted physical goods or listed entities. Now they’re hitting the digital layer — the actual models and services. That shift makes compliance trickier for any company that operates across borders. For decentralized platforms, which often lack a central point to enforce restrictions, the challenge is even steeper.
The pull of Fable 5 and Mythos 5 is a concrete example of how quickly access can disappear. Users who relied on those tools now have to look elsewhere, and the broader AI community is watching to see which product gets yanked next.
What’s next for compliance
Anthropic has not detailed whether it will appeal the directive or try to re-release modified versions that comply. The company hasn’t said if the controls apply to other products in its lineup. For now, the precedent stands: software-level export controls are real, they’re enforceable, and they can cut off access overnight.
