U.S. government approval now gates AI model releases for OpenAI, Anthropic

What happened

The U.S. government has begun controlling which AI models get released. Anthropic's Fable and Mythos models were pulled by the government two weeks ago, and OpenAI's GPT 5.6 is now headed for limited preview with government approving the release "customer by customer" until a general release can be approved.

Why it matters

The approval process risks slowing the release of expensive new AI systems at a time when labs are struggling to improve their bottom lines. Mythos has already been in preview for months with no indication of general release coming soon, and even a few weeks in review could significantly limit the economic value of a costly new system. If model development slows as a result, it could put a chill on the ongoing data center buildout that the industry depends on.

What to watch

The U.S. government lacks the expertise and capacity to properly test these models before release, and there is no clear articulation of what risks regulators are actually concerned about. The article notes that both OpenAI and Anthropic now face the same problems, and solving them will require the industry to work together rather than using regulation as competitive advantage.