The Trump administration has asked OpenAI to restrict the launch of its newest model, GPT-5.6, releasing it only to a small group of government-approved partners before any wider rollout. The request, first reported by The Information, was driven by security and safety concerns tied to the model's advanced capabilities, according to coverage from TechCrunch, Axios, Bloomberg, CNN and The Verge.
A source familiar with the matter told outlets including MSN that the White House wanted access limited until a cybersecurity review is complete. OpenAI, which has billed GPT-5.6 as "our strongest model yet," will reportedly stagger the rollout so that only "trusted partners" can use it for now, per The Stack and Engadget.
The Verge reported that OpenAI will delay the model after the administration's request. Some coverage referred to the system as "GPT-5.6 Sol," positioned as a rival to Claude Mythos; the-decoder reported that OpenAI considers the government access rules "unsustainable." SiliconANGLE noted the company is staggering the release for government vetting while eyeing a 2027 IPO.
The episode marks a notable shift from earlier industry practice. As Engadget put it, "So much for voluntary review" — a nod to the fact that frontier AI safety checks have largely been self-imposed until now.
Why it matters: This is one of the clearest signs yet that the U.S. government is moving from suggesting AI safeguards to actively shaping when and how the most capable models reach the public.