Who decides when AI is too dangerous?

Today's episode centers on who decides the line between safe and dangerous AI after Anthropic's Fable 5 triggered regulatory drama. On Friday, the U.S. government imposed export controls on Fable and the Mythos model it is based on, restricting access for foreign nationals, including Anthropic staff in the United States. Anthropic then pulled Fable and Mythos offline to comply with the order. As of Tuesday, Hayden Field noted in The Verge timeline that the model remained unavailable, with Claude signaling that Fable 5 is currently unavailable. The disruption underscores broader questions about AI safety governance and how aggressively the government should police cutting‑edge models. Anthropic had argued that AI could become dangerous and that regulation should come sooner rather than later, so the standoff marks a test of that stance. The move also highlights the fragility of private controls when export rules bite across borders. Hayden Field outlined a weekend scramble and ongoing fallout that could influence future policy decisions. It is unclear how the dispute will be resolved or what precedent it sets for the U.S. regulatory regime. Observers will watch whether the administration moves toward a formal safety framework or relies on ad hoc actions. The episode underscores the ongoing tension between national security interests and rapid AI deployment. The outcome may reshape access to powerful models and influence how firms navigate cross-border collaboration.