OpenAI launched its GPT-5.6 series on June 26, gating access to Sol, Terra, and Luna behind what the company called “trusted partners whose participation has been shared with the government.” The restriction came at the Trump administration’s request, making this the first preemptive White House constraint on a frontier model launch, and one delivered without statutory authority or a designated lead agency.

The mechanics are improvised. Trump signed a June executive order asking labs to voluntarily submit advanced models for a 30-day pre-release review, but the implementing framework doesn’t yet exist. In its absence, agencies have been freelancing. The White House handled OpenAI directly. Commerce, separately, forced Anthropic’s Mythos offline through an export-control order, and Anthropic also disabled Fable to comply. A source told CNN that Sol is considered “on par” with Mythos, which suggests the asymmetry isn’t about capability tiers so much as which agency picked up the phone first.

Brad Carson of Public First put the structural problem bluntly: “Right now, you have an ad hoc, personalized, opaque, possibly lawless approach.”

OpenAI, for its part, is complying while flagging the precedent. The company said it previewed the models to the government before launch and is working with the administration on a “repeatable process for future model releases,” with general availability for Sol, Terra, and Luna expected in the coming weeks. It also drew a line: “We don’t believe this kind of government access process should become the long-term default.”

That sentence is the entire story. A frontier lab is publicly negotiating the terms of its own regulation in a blog post, because there’s no other venue in which to do it.

Sources

Sources