OpenAI said on Friday it is restricting the release of its newest artificial intelligence model at the request of President Donald Trump's administration, while the model undergoes a cybersecurity review. According to reporting carried by ABC News and other outlets, the new ChatGPT product will initially be available only to customers approved by the U.S. government.
The model is GPT-5.6, released in a limited rollout. Coverage from Livemint and CNBC-TV18 describes multiple variants, including ones called Sol and Terra, and notes the model ships with stronger cyber protections.
OpenAI is not alone. Several outlets, including the Ottumwa Courier and others via Google News, report that rival Anthropic is taking a similar step, limiting its new AI models to government-approved customers during a cybersecurity review.
The restriction reflects how capable these systems have become. As NDTV framed it, the model was powerful enough that OpenAI briefed the U.S. government before release.
OpenAI signaled it is uneasy about the arrangement. "We don't believe this kind of government access process should become the long-term default," the company said, according to MSN's report. "It keeps the best tools from users, developers, enterprises, cyber defenders, and global" users. The company stressed that such restrictions should not become the norm and pushed for wider access soon.
Separately, OpenAI named Prabhjeet Singh as its Managing Director for India, per Livemint.
Why it matters: This is an early, unusual case of the U.S. government acting as a gatekeeper before powerful AI reaches the public — a precedent that could reshape who gets access to cutting-edge tools, and when.