The United States government has asked OpenAI to limit access to its upcoming GPT-5.6 artificial intelligence model to a small group of government-approved partners, due to security risks associated with powerful AI systems.
According to CNN, the request follows the government’s recent order requiring Anthropic to withdraw its advanced Mythos and Fable AI models after officials pointed to potential cybersecurity risks.
A source familiar with the matter said the White House considers GPT-5.6 to be comparable to Anthropic’s Mythos in capability. Rather than stopping the launch entirely, the government asked OpenAI to introduce the model through a controlled rollout.
OpenAI agreed to the request. In an internal memo sent to staff on Thursday, CEO Sam Altman said access to GPT-5.6 would be approved “customer by customer” during the initial release.
“We’ve made clear to the U.S. government that this is not our preferred long-term model, and will work with them and others in industry to achieve a more sustainable approach for future releases,” Altman said in the memo.
A White House official told CNN that the administration is continuing to work with leading AI companies to develop shared approaches for managing the challenges created by increasingly advanced AI models. OpenAI declined to comment.
The request has also drawn attention to the lack of a clear regulatory framework for advanced AI in the United States.
Earlier this month, President Donald Trump signed an executive order asking companies developing frontier AI models to voluntarily submit them for government review 30 days before public release. However, officials are yet to establish how the process will work.
The current situation has also created uncertainty within the AI industry. While the White House requested OpenAI’s limited rollout, the Commerce Department issued the export control order that forced Anthropic to suspend access to Mythos and Fable, leaving companies without a single, consistent regulatory process.
Experts say government oversight is important because advanced AI models can identify software vulnerabilities and simulate sophisticated cyberattacks. However, they also warn that regulation should be transparent and predictable.
“The Fable episode shows the need for clear regulations. Right now, you have an ad hoc, personalised, opaque, possibly lawless approach,” Brad Carson, head of Public First, a bipartisan pro-AI safety organisation, told CNN.
“It is certainly appropriate for the government to recall dangerous products, including AI models, but it has to be done in a way consistent with transparency and basic fairness.”
GPT-5.6 is expected to undergo testing with a limited number of enterprise partners before OpenAI decides on a wider public release.
