Trump’s 30-Day AI Review Order Drives OpenAI to Cap GPT-5.6 Release

For legal and regulatory professionals, this marks a pivotal test of the president’s power to impose de facto pre‑market approval requirements on emerging technologies without explicit statutory backing. The White House’s leveraging of a voluntary framework—backed by the threat of future mandatory action—creates a compliance landscape where AI labs must navigate ambiguous legal boundaries. This move could set a precedent for how far executive orders can reach in areas traditionally governed by congressional legislation, with implications for constitutional law, administrative procedure, and international competitiveness.

OpenAI’s decision to restrict the release of its newest AI model, GPT-5.6, to a small, administration-approved group of U.S. companies and organizations marks a significant escalation in the executive branch’s efforts to govern frontier artificial intelligence. According to multiple sources, the move came at the direct request of the Trump White House after consultations with the Office of Science and Technology Policy and the Office of the National Cyber Director. This is not an isolated incident but part of a broader pattern: earlier in June 2026, President Trump signed an executive order aimed at managing the release of powerful AI models with advanced hacking capabilities. The order was itself a response to the restricted launches of Anthropic’s Claude Mythos and OpenAI’s own GPT-5.5-Cyber, both of which were made available only to trusted researchers and cybersecurity firms due to concerns over potential cyberattacks.

“

OpenAI’s decision to restrict the release of its newest AI model, GPT-5.6, to a small, administration-approved group of U.S.

The key mechanism in the executive order is a request—not a mandate—for companies to voluntarily submit new models for government review at least 30 days before planned public release. By acceding to the request for GPT-5.6, OpenAI is effectively setting a precedent that the administration can exert substantial control over AI deployment without formal rulemaking or explicit congressional authorization. This raises fundamental questions about the scope of presidential authority in technology regulation, the legality of voluntary but pressure‑driven compliance, and the potential for de facto prior restraint on innovation.

From a governance standpoint, the White House is leveraging a soft‑power approach that blurs the line between voluntary cooperation and regulatory coercion. The leaked memo from OpenAI CEO Sam Altman, which described the limited release as not the “preferred” method but expressed hope for a “sustainable” government dialogue, underscores the delicate balancing act companies face. On the one hand, they want to maintain good relationships with the administration and avoid the reputational risk of releasing a model that might later be blamed for a catastrophic cyber event. On the other, the pre‑review process introduces uncertainty, potential delays, and a competitive disadvantage if rivals from other countries—or even domestic competitors who refuse to cooperate—move faster.