OpenAI's next AI model, reportedly called GPT-5.6, will not get the wide public launch the company planned. At the request of the U.S. government, the model will initially be available only to select partners.
According to The Decoder, access will be granted on a "customer by customer" basis, meaning the rollout now effectively requires U.S. government approval for each customer. The Decoder reports that OpenAI CEO Sam Altman has described this arrangement as not a "preferred long term model," suggesting the company sees it as a temporary constraint rather than its intended way of doing business.
The Financial Express reports that the administration asked OpenAI to "stagger" the GPT-5.6 release, framing the move as a second attempt by the U.S. to control the global AI race. It adds that OpenAI is reportedly debating whether to delay the model's public debut until next year.
MSN reports that the administration is citing national security concerns and wants to evaluate the model's potential before broader release.
Multiple outlets tie the decision to a recent confrontation involving rival lab Anthropic. The Decoder refers to a "forced takedown" of Anthropic's model Fable, and Benzinga frames the GPT-5.6 limits as coming "after the Anthropic showdown." Yellow.com reports that OpenAI was preparing to challenge Anthropic with GPT-5.6 before the White House intervened.
The specific facts here come from press reports rather than official confirmation, and the sources do not detail exactly how the approval process will work or which customers qualify.
Why it matters: If the U.S. government begins deciding, customer by customer, who can access the most advanced AI models, it marks a significant shift toward treating frontier AI like a controlled strategic technology rather than an ordinary commercial product.