OpenAI is delaying the full public rollout of its latest AI model, GPT-5.6, at the request of the United States government.
The company announced it will initially limit access to a small group of vetted partners, with their details shared with authorities. This step follows growing concern in Washington over national security risks posed by powerful AI systems.
By securing early access, U.S. officials aim to identify potential threats, from cyberattacks to military misuse, before wide deployment. The move is part of a broader effort to establish guardrails.
This limited release aligns with a recent executive order creating a voluntary framework for AI developers. The framework offers the government access to advanced models for up to 30 days before their release to trusted partners.
OpenAI described the move as a temporary measure while working with the administration on a permanent process. The company cautioned, however, that such oversight should not become the permanent standard, expressing concern it could restrict access for beneficial users.
The new model lineup includes the advanced GPT-5.6 Sol, the mid-tier Terra, and the lower-cost Luna.