OpenAI’s next model rollout may look very different from earlier launches, as the White House has reportedly asked the company to initially limit GPT-5.6 access to select partners while federal officials assess security risks.
TL;DR
- OpenAI reportedly plans a limited GPT-5.6 preview for select partners.
- The request came amid U.S. national security and cybersecurity concerns.
- A broader release could follow a couple of weeks later, depending on review outcomes.
OpenAI is reportedly preparing to release GPT-5.6 in a limited preview rather than making it broadly available from day one.
According to Reuters, which cited a source familiar with the matter, the Trump administration asked OpenAI to stagger the release of its new model over security concerns. The Information first reported the development, while TechCrunch and Axios also reported that the initial preview would be restricted to a small group of partners.
At a meeting this week, OpenAI CEO Sam Altman reportedly told staff that the government would be approving access customer by customer during the preview period. He also reportedly said that if the limited release goes well, OpenAI hopes to follow with a broader release a couple of weeks later.
What makes this notable is that it appears to be one of the clearest examples yet of the U.S. government stepping in before a major American AI model reaches the broader market.
Axios reported that the White House’s Office of the National Cyber Director and Office of Science and Technology Policy asked OpenAI to limit the rollout as the administration builds a framework for testing and evaluating the security of new models. It also reported that Altman discussed GPT-5.6 with Commerce Secretary Howard Lutnick, who wanted relevant government bodies to test and approve the model.
The move follows a June 2, 2026 executive order titled Promoting Advanced Artificial Intelligence Innovation and Security, which directed federal agencies to develop a voluntary framework for frontier model access and testing. The order says developers may provide the government with access to covered frontier models for up to 30 days before release to other trusted partners.
However, the same order also says it should not be read as creating mandatory licensing, preclearance, or permitting requirements for new AI model releases. That distinction matters, as the current OpenAI situation appears to sit somewhere between voluntary coordination and government-requested restraint.
As per Axios, Altman told employees, “We’ve made clear to the U.S. government that this is not our preferred long term model,” while adding that OpenAI would work with the government and industry on a more sustainable approach for future releases.
For OpenAI, the timing is important.
The company recently published a Frontier Governance Framework, saying its safety and security practices cover risk assessment and mitigation across cyber offense, chemical, biological, radiological and nuclear risks, harmful manipulation, loss of control, incident response, external expert input and model reporting.
OpenAI also said its GPT-5.5 model was treated as High for biological, chemical and cybersecurity capabilities under its Preparedness Framework, while adding that the model went through preparedness evaluations, targeted advanced biology and cybersecurity testing, and external expert review before release.
This helps explain why GPT-5.6 is drawing closer federal attention, especially if it represents a further jump in cyber or agentic capabilities.
The broader AI sector is already moving toward controlled access for powerful models. Anthropic, for example, launched Project Glasswing to give selected partners access to Claude Mythos Preview for finding and fixing vulnerabilities in foundational systems. Anthropic later said it was expanding the program to around 150 new organizations across more than 15 countries, each subject to security requirements.
For now, GPT-5.6 is not being publicly framed as delayed indefinitely. Instead, the reported plan suggests a short, controlled preview period designed to satisfy national security concerns before a wider release.
Still, the episode could set a precedent for how frontier AI models are launched in the U.S. going forward. The AI race may still be about speed, capability and market share, but this latest move shows safety reviews and federal coordination are becoming part of the release calendar.
