The Trump administration has asked OpenAI to limit the initial release of its upcoming GPT-5.6 artificial intelligence model to a small group of government-approved partners over security concerns.
Washington has intensified its focus on the security implications of increasingly powerful AI systems, claiming that advanced models could be exploited by hostile nation-states, cybercriminals or other malicious actors.
According to Axios, the White House's Office of the National Cyber Director and the Office of Science and Technology Policy asked OpenAI to limit GPT-5.6's rollout while the administration develops a framework for testing and evaluating the security of advanced AI models.
OpenAI Chief Executive Sam Altman informed employees that GPT-5.6 would initially receive a limited rollout and expressed hope that a broader release could follow within a couple of weeks, The Information reported. Altman also said the company had informed the U.S. government that the arrangement was not its preferred long-term approach and that it would continue working with officials and industry partners on a more sustainable framework for future model releases.
Axios also reported that Commerce Secretary Howard Lutnick discussed GPT-5.6 with Altman on Wednesday. According to the report, Lutnick wanted to ensure all relevant parts of the federal government had reviewed and approved the model before any wider release. A source cited by the publication said officials viewed GPT-5.6 as having capabilities comparable to Anthropic's Mythos model, prompting additional scrutiny rather than signaling a broader shift toward tighter regulation.
The administration's request follows President Donald Trump's executive order signed earlier this month establishing a voluntary framework for government testing of advanced AI models before public release. The order directs federal agencies to develop cybersecurity standards and create a process for AI developers to voluntarily share their most capable models with the government for security evaluations.
The latest development also comes less than two weeks after Anthropic disabled access to its flagship Fable 5 and Mythos 5 models following a U.S. government directive restricting access for foreign nationals over national security concerns. Anthropic said the government believed the models could potentially be "jailbroken" to identify software vulnerabilities, although the company said it had not received detailed explanations supporting the concerns, Reuters reported.
The broader backdrop is an increasingly competitive race among leading AI developers to release more capable models while governments weigh the security risks associated with frontier AI. U.S. officials have repeatedly expressed concerns that advanced AI systems could assist cyberattacks, accelerate vulnerability discovery or be exploited by foreign adversaries if adequate safeguards are not in place.
The administration plans to allow only about two dozen organizations to receive early access to GPT-5.6, with the government reviewing potential users during the preview period, The Financial Times reported. The report said the approach is intended to give federal agencies additional time to evaluate the model before a wider rollout.
The administration has maintained that its AI security efforts are intended to strengthen protections without slowing U.S. innovation.
