OpenAI has released its newest model, GPT-5.6, but most people won't be able to use it. According to reporting aggregated from Nextgov/FCW, Broadband Breakfast and the Marietta Daily Journal, the company is limiting access to a small set of "trusted" or "Trump-approved" customers, at the request of the US government, while a cybersecurity review is underway. The release is being offered in the United States only.
PCWorld summed up the mood with its headline that the powerful new models have arrived "but not for you." Stocktwits reported that the access limits came specifically at the US government's request, framing the chosen recipients as "trusted partners." Computerworld described it more bluntly as the US telling OpenAI to restrict access to its most powerful AI model.
The move follows a similar one aimed at a rival. According to Moneycontrol, the GPT-5.6 limits come about two weeks after the US government surprised Silicon Valley by ordering Anthropic to ban all foreign nationals from accessing its Fable 5 and Mythos 5 models. Computerworld noted that US authorities are growing "twitchy" about so-called frontier AI systems, and Digit reported that OpenAI was expected to follow Anthropic's lead in restricting early access.
Not everyone agrees the caution is warranted. In an opinion piece, The Washington Post argued that having federal officials decide who gets GPT access reflects a misreading of the actual threats posed by AI.
Why it matters: For the first time, the US government is effectively deciding who can use the most capable commercial AI systems—a shift that turns access to cutting-edge models into a question of national security and government approval rather than open availability.