OpenAI Restricts New AI Models to 'Trusted Partners' at US Government's Request

OpenAI has restricted its newest AI models to a closed group of 'trusted partners', acting at the direct request of the US government — a move that marks a striking new frontier in state oversight of artificial intelligence development. The decision means that, unlike previous OpenAI releases which were made broadly available, the latest models will not be accessible to the general public or international developers without specific clearance.

The US government's involvement goes further than previously reported. According to CNBC, OpenAI briefed American authorities before the release of its new AI, with the government apparently satisfied that wider immediate access posed risks that needed to be managed. The arrangement effectively makes the US administration a gatekeeper in OpenAI's commercial release process — a development with no real precedent in the tech industry at this scale.

The move comes amid a wider pattern of Washington tightening its grip on frontier AI. Within the same news cycle, Anthropic received conditional approval for its own Mythos 5 model, also under strict government-imposed limitations. Together, the two developments suggest the US is now treating the most powerful AI systems as strategic assets requiring government sign-off before deployment — much like advanced semiconductors or military technology.

For India, the implications are significant. Indian AI companies, academic institutions, and the growing cohort of startups building on top of OpenAI's models could find themselves outside the trusted-partner circle, facing delays or restricted access to the most capable tools. As India accelerates its own AI ambitions, the gatekeeping decisions made in Washington are poised to shape what the country's developers can build — and how quickly.