Trump Administration Asks OpenAI to Limit Release of GPT-5.6 Model The Trump administration’s move to slow the rollout of OpenAI’s next flagship model, GPT-5.6, has turned a long‑running debate over AI safety into a live test of how far Washington can go in steering private AI development.

In early June, President Trump signed an AI security executive order directing agencies to build processes for testing powerful models before public release, laying the groundwork for closer oversight. As officials assessed OpenAI’s forthcoming GPT-5.6, the White House’s Office of the National Cyber Director and the Office of Science and Technology Policy pressed for a restricted launch while they develop a broader security framework.

By June 25, multiple outlets reported that the administration had formally asked OpenAI to limit GPT-5.6’s initial release to a “small set of government‑approved partners,” marking the first time the U.S. government has preemptively requested an American AI company to restrict a model before launch. TechCrunch described the shift as the White House “asking OpenAI to slow roll the release of its new model over safety concerns.”

Inside OpenAI, CEO Sam Altman told employees the company would comply by offering a limited preview to select enterprise customers, with the government “approving access customer by customer” and a possible wider rollout “a couple of weeks later” if the trial goes well. Altman stressed in a memo that this is “not our preferred long term model” and that OpenAI wants a “more sustainable approach for future releases.”

The Trump administration frames the move as narrowly targeted at models with “Mythos‑like” capabilities, referencing rival Anthropic’s restricted Claude Mythos system, which the company itself kept “under wraps” over fears of misuse. Yet critics see an uneven, increasingly heavy‑handed posture: Anthropic recently faced an export‑control directive barring “foreign nationals” from accessing its Mythos 5 and Fable 5 models, while OpenAI has so far been granted a more flexible, case‑by‑case preview regime.

Online, the episode revives a longer history of contentious slowdown decisions in AI. One viral post recalls that “Dario once delayed the release of GPT‑2 back at OpenAI, claiming it was too dangerous,” underscoring how arguments over when a model is ‘too powerful’ have shadowed frontier AI for years.

Continue reading https://foxvector.com/stories/019f0263-a9da-05ee-72af-1e0a65d7cf87