On June 2, 2026, President Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security," directing several national security and civilian agencies to ramp up scrutiny of cutting-edge AI models. The order asks AI developers to submit certain cutting-edge models to a group of federal agencies for a voluntary review 30 days before they are made public, with the White House intending to use that time to identify and remediate security weaknesses that such models could exploit. The order instructs the Treasury Department, the National Security Agency, the Cybersecurity and Infrastructure Security Agency, the National Institute of Standards and Technology, and other federal bodies to establish the new framework within 60 days. Crucially, the order is voluntary, not mandatory, and does not create a licensing regime or a preclearance requirement for AI development.
The first real test of that line between voluntary and mandatory arrived almost immediately. Due to government restrictions levied by the Trump administration, GPT-5.6 was first released as a limited preview only on June 26, 2026, for a "small group of trusted partners," despite a planned public rollout in the same month. Commerce Secretary Howard Lutnick met with OpenAI CEO Sam Altman to discuss GPT-5.6, with his goal being to verify that agencies across the government had reviewed the model and signed off on it. The Department of Commerce's Center for AI Standards and Innovation conducted additional tests, with OpenAI sending technical experts to Washington so they could immediately address potential questions and concerns. OpenAI released the GPT-5.6 family publicly on July 9, 2026, following the limited preview that had begun on June 26. The family includes Sol, the new flagship; Terra, a balanced model for everyday work; and Luna, the most cost-efficient model, with Sol setting new benchmarks across coding, knowledge work, cybersecurity, and science.
The Anthropic episode, unfolding in parallel, illustrated what happens when the informal framework is bypassed entirely. The U.S. government ordered Anthropic to take down its new AI models, Fable 5 and Mythos 5, just three days after their public launch on June 12, 2026; citing national security and export-control powers, the Commerce Department mandated the suspension of access for all foreign nationals, leading to a global shutdown, after reports of a jailbreak bypassing safety guardrails. The order required Anthropic to halt access for "any foreign national," including the company's own non-citizen employees, and Anthropic, citing the technical impossibility of filtering users by nationality across dozens of global cloud platforms in real time, enforced an all-customer, all-integration universal shutdown. The U.S. Department of Commerce ultimately lifted the export controls on Fable 5 and Mythos 5, ending an 18-day restriction.
Skeptics warn that the practical consequences of the new regime extend well beyond any single product launch. Dean Ball, a former White House AI adviser who is soon to join OpenAI, argues that the ostensibly "voluntary" mechanism has evolved into a de facto involuntary licensing regime, leading to heavy-handed government restrictions on model releases. The restricted GPT-5.6 rollout fit within, but also exceeded, the broader oversight structure, and where the framework stopped at voluntary participation, the GPT-5.6 situation escalated into a state-curated access list, a concession OpenAI made only after federal officials pressed it to delay. The crackdown on Anthropic was gaining criticism from a number of tech executives and investors, in part because it handed valuable time to Chinese open-source developers that are trying to catch up. Companies that rely on frontier AI models but do not build them face a new dependency risk: the 30-day review window could delay access to new capabilities, with some real competitive consequences for businesses whose differentiation depends on early adoption.
The paperwork says "voluntary," but OpenAI flew engineers to Washington and waited 13 days for a nod that the White House insists it never gave.