According to a CNN report citing an internal source, the U.S. government has asked OpenAI to restrict access to its soon‑to‑be‑released GPT‑5.6 model to a small number of state‑approved partners. Officials said the request reflects concern about the model’s advanced capabilities.
Background to the request
The request followed export controls the U.S. imposed on Anthropic, which led that company to withdraw its most advanced systems, Mythos and Fable. CNN’s sources say that OpenAI and U.S. officials consider GPT‑5.6 to be "equivalent" in capability to Mythos.
The move comes amid a broader but still-evolving federal approach to next‑generation AI. Earlier this month, President Donald Trump signed an order asking companies building advanced models to voluntarily submit their systems for government review 30 days before public release, but the practical details and implementing procedures of that directive have not yet been settled.
How the access limitation will work
Citing an internal memo reported first by The Information, CEO Sam Altman said the government will approve access on a case‑by‑case, client‑by‑client basis. OpenAI accepted the limited rollout because it creates an interim path to market while a unified federal regulatory framework for these new models is still lacking.
Coordination challenges and criticism
The reporting highlights uncertainty among industry participants about which agency will coordinate processes during the transition. The OpenAI request came directly from the White House, while the export controls on Anthropic were imposed by the U.S. Department of Commerce, underscoring a lack of a single coordinating authority.
Experts agree that the government should be involved in negotiations over AI safety, particularly for national‑security risks. Brad Carson, head of the bipartisan Public First supercommittee, argued that the current ad‑hoc, opaque and legally uncertain approach should be replaced by predictable regulation that upholds transparency and fairness.
Government and industry response
A White House official confirmed ongoing close cooperation with leading AI developers to address challenges arising from the technology’s scaling. OpenAI declined to comment on the CNN report.
The U.S. request to limit GPT‑5.6’s availability illustrates a cautious, targeted approach to managing risks from powerful new AI models while broader, longer‑term regulatory solutions remain under development.
