
OpenAI is reportedly planning a limited rollout of its upcoming ChatGPT 5.6 model, with early access restricted to customers approved by the US government before a broader public release.
According to a report, OpenAI CEO Sam Altman informed employees in an internal memo that federal authorities will approve access “customer by customer” during an initial preview period. The report adds that a wider release is expected to follow a few weeks later.
The move comes after US President Donald Trump signed an executive order earlier this month encouraging AI companies to submit their most advanced models for voluntary federal review before public launch. The gover
Altman reportedly acknowledged that OpenAI does not view the current arrangement as a long-term solution.
According to the memo, the company has informed the US government that it will continue working with officials and industry partners to establish a more sustainable process for future AI releases.
The report by The Information says multiple US agencies are involved in shaping the review process, including the Office of the National Cyber Director, the Office of Science and Technology Policy, and the Department of Commerce. However, government officials have not publicly commented on the reported approval mechanism.
The planned rollout also follows recent restrictions imposed on rival AI company Anthropic. Earlier this month, Anthropic reportedly suspended access to two of its latest models after receiving a federal directive that blocked availability to foreign nationals, citing security concerns.
The developments have raised fresh questions about how voluntary the US government’s AI review process will be in practice. While the executive order describes the reviews as voluntary, recent actions involving both OpenAI and Anthropic suggest federal oversight of advanced AI models may become more active before public deployment.
OpenAI has not officially announced the release timeline for ChatGPT 5.6 or confirmed the reported government approval process. If implemented, it would mark one of the first major AI launches in which access is controlled through a government-supervised preview before wider availability.