U.S. government to approve GPT-5.6 access on a per-customer basis
OpenAI will release GPT-5.6 in restricted preview, with U.S. government officials approving each customer individually before granting access, CEO Sam Altman said.
The Trump administration has asked OpenAI to withhold broad access to GPT-5.6, requiring federal approval for each client during an extended preview period. CEO Sam Altman confirmed the arrangement, describing the vetting process as "client-by-client" approval by government officials citing national security concerns.
GPT-5.6 will launch in limited preview rather than a general rollout. During the preview window, U.S. government officials will review and approve individual customer requests before OpenAI grants access. Altman did not specify how long the preview period would last or what criteria officials would use to approve or deny access. The arrangement effectively introduces de facto licensing for frontier model deployment, even though no formal regulatory framework yet exists.
The move marks the first time a major U.S. AI lab has agreed to government pre-approval of customer access for a new model release. OpenAI's previous GPT-4 and GPT-4 Turbo launches went live to all API customers and ChatGPT subscribers without a vetting layer. Whether GPT-5.6 will eventually reach the general public remains unclear — Altman's statement left open the possibility that the preview could extend indefinitely.
