U.S. control over AI model releases threatens the industry

The U.S. government is increasingly determining which AI models can reach users. After pulling Anthropic’s Fable and Mythos, OpenAI’s GPT 5.6 appears headed to a limited, customer-by-customer preview rather than an immediate general release. If that preview lasts longer than a few weeks—Mythos has already been in preview for months—it could cut into the economic upside of an expensive new system at a moment when labs are trying to shore up their finances.

OpenAI and Anthropic now face the same constraints and the same potential fallout if regulatory reviews stall. Industry arguments have fixated on who brought this on, ranging from claims of regulatory capture to political maneuvering, but the underlying problem is broader: a fragmented government approval process imposes costs across the sector, and there is no fix that helps one lab without helping the others.

Creating a release process that makes sense is the immediate task.

United States

u.s. government, ai models, model releases, anthropic, openai, fable, mythos, gpt 5.6, regulatory reviews, limited preview