OpenAI delays GPT-5.6 release after Trump administration mandates case-by-case customer approval

OpenAI will release its next model, GPT-5.6, in limited preview form only, with the Trump administration approving customer access on a case-by-case basis, according to The Information.

CEO Sam Altman informed employees during a Wednesday company Q&A that OpenAI would grant access to "a small group of enterprise customers" in compliance with a federal government request citing security concerns. The administration will individually approve each customer during the preview period.

Uneven regulatory approach

The arrangement differs significantly from treatment of OpenAI's competitor Anthropic, which received an ultimatum earlier this month requiring it to suspend access to its Mythos 5 and Fable 5 models. The administration issued an export control directive prohibiting "foreign nationals" from accessing Anthropic's technology, including the company's own non-US citizen employees.

The disparate regulatory treatment has raised questions about consistency in AI oversight. The Trump administration previously promised a "speed wins" approach to AI technology and pledged to encourage American AI exports.

Limited details on GPT-5.6

OpenAI has not disclosed technical specifications for GPT-5.6, including context window size, pricing, or benchmark performance. The model's capabilities and release timeline beyond the government-mandated preview period remain unclear.

The export control measures represent a shift from the administration's earlier AI policy statements. Industry observers note the case-by-case approval process could significantly slow enterprise adoption compared to standard model releases.

What this means

The Trump administration is implementing AI export controls unevenly across companies, with OpenAI receiving more favorable treatment than Anthropic despite operating in the same space. The case-by-case approval requirement will create deployment delays and operational complexity for enterprise customers, particularly those with international operations. This marks the first time a major US AI lab has been required to gate model access through direct government approval, setting a precedent that could extend to future releases across the industry.