OpenAI Unveils GPT-5.6 With Staggered Release at US Government's Request

OpenAI Unveils GPT-5.6 With Staggered Release at US Government’s Request OpenAI’s launch of its next-generation GPT‑5.6 models has become a real-time test of how much control Washington will exert over frontier AI, as the company rolls out powerful systems under unprecedented, government‑managed restrictions.

On June 2, President Donald Trump signed a narrower AI executive order that asks leading labs to “voluntarily submit their most advanced models for government review up to 30 days before release,” a move critics say has created a “de facto involuntary licensing regime” for top‑end AI. In mid‑June, the administration used separate export controls to force rival Anthropic to revoke access to its Mythos and Fable 5 models, signaling a tougher line on systems with advanced cyber capabilities.

Behind the scenes, U.S. agencies including the Office of the National Cyber Director, the Office of Science and Technology Policy, Treasury and Commerce began working with OpenAI on GPT‑5.6, asking the company to limit distribution while they build a framework for testing and evaluating new models’ security. Axios reported this was “the first time the U.S. government has preemptively asked an American AI company to restrict the launch of a model before release.”

On June 25, The Information and others revealed that CEO Sam Altman had told staff GPT‑5.6 would debut only in a limited preview, with the government “approving access customer by customer” during the initial period. The Verge and TechCrunch confirmed that the Trump administration had requested OpenAI “slow roll the release of its new model over safety concerns.”

Less than 24 hours later, on June 26, OpenAI formally unveiled GPT‑5.6 — a family of three models, Sol, Terra and Luna — emphasizing strengths in coding, cybersecurity and biology and describing Sol as the flagship. A system card framed the launch as coming with OpenAI’s “most robust” safeguards yet, designed to deliver the models safely at scale.

Yet access was sharply constrained. OpenAI said it was starting with “a limited preview for a small group of trusted partners whose participation has been shared with the government,” in line with its Defense Department agreement. Axios reported around 20 companies had been approved, with Washington now treating frontier models as products needing review before broad release. The Next Web noted Sol’s rollout to “roughly 20 partners whose names were individually approved by the US government,” calling it the first frontier model launched under a government‑managed access list.

Altman acknowledged the trade‑off publicly, praising Sol as “a smart, efficient, and a significant step forward” but adding that “at the request of the US government, it is launching today in limited preview.” OpenAI argued in a blog post that “we don’t believe this kind of government access process should become the long‑term default,” warning it “keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.”

From industry and policy analysts, the move is seen as part of a broader power shift. TechCrunch wrote that “the U.S. government is set to take an awful lot of control over which AI models get released,” putting OpenAI and Anthropic “in the same exact position” and risking slower development and chilled data‑center investment if approval drags out. Former White House adviser Dean Ball has argued the emerging process lacks clear safety standards, raising the prospect of “endless launch delays” that could also give China an edge.

Other AI leaders are more supportive of tighter oversight of proprietary frontier APIs, while sparing open‑source models. Hugging Face CEO Clément Delangue wrote that “it’s quite rational to regulate frontier API models, especially to get more transparency for the government, without regulating open‑source AI,” arguing that the “most dangerous AI systems right now aren’t open models” but large proprietary APIs. Meta’s Yann LeCun amplified the view by resharing Delangue’s post.

OpenAI, for its part, stresses the shift is temporary. Axios and Business Insider both report the company expects a broader release in the “coming weeks” if the preview raises no new concerns, and that it has told the administration this “is not our preferred long term model” for launches. But with GPT‑5.6’s debut now bound up with the Trump administration’s new AI order, the episode is likely to shape how future frontier systems — from any lab — make it from research to public hands.

Continue reading https://foxvector.com/stories/019f13c8-67ad-1afc-7311-2314f763efd7