EO 14409 disclaims any federal preclearance for frontier AI models. The procurement, contracting, and partnership levers around it do the work the disclaimer disclaims.
President Trump signed Executive Order 14409, "Promoting Advanced Artificial Intelligence Innovation and Security," in June 2026. The order asks frontier AI companies — the labs building the largest, most capable models — to share those models with the federal government before releasing them to the public. The headline ask has gotten most of the attention. The single most consequential sentence in the order is the one saying what the order refuses to do.
Section 3(c) of EO 14409 explicitly disclaims the creation of "any mandatory governmental licensing, preclearance, or permitting requirement" for new AI models, including frontier models. That disclaimer is where the real federal-oversight question lives. If the order is read at face value, the pre-release request is a courtesy. Read against the levers the rest of the order activates, the disclaimer is the clause doing the most work, and the clause most likely to be tested in court or in practice.
The Council on Foreign Relations' assessment of the order reaches the same point from the policy-review side. The procurement, contracting, and partnership mechanisms the order actually deploys to coax compliance are doing the work the disclaimer disclaims. The order does not compel disclosure. It directs federal agencies to use their existing authorities — federal grants, procurement preferences, test-bed partnerships — to make voluntary disclosure the path of least resistance for any lab that wants to work with Washington. That posture is different from a permission regime, and more durable, because it does not require new statutory authority to enforce.
The order has resurfaced in public conversation this week after NPR's Michel Martin interviewed Diego Munhoz, a tech reporter at Punchbowl News, about how the order is being read inside the capital. Spokane Public Radio carried the segment on July 31. The Guardian's coverage of the order as "seeking early access to new AI releases" captures the same voluntary-review posture the text itself uses. Neither outlet disputes the order's framing; both treat the disclaimer as the load-bearing clause.
A pre-release review window could give regulators an early look at safety and capability risks before the public encounters a model. It also raises real questions about intellectual property, the competitive burden on US labs, and how much authority the executive branch should hold over commercial AI. Both sides of that tension are defensible. Regulators have a legitimate interest in seeing frontier capabilities before they ship, especially as the systems move into critical infrastructure and defense work. Labs have a legitimate interest in not being treated as applicants for permission to release their own models, and in keeping proprietary weights out of any government system that does not have an airtight security story.
The next test is not the language of the order. It is whether federal agencies use their procurement, contracting, and partnership levers to make voluntary disclosure the de facto price of doing business with Washington. The order was published in the Federal Register on June 5, 2026; the CFR assessment is the clearest public reading of how Section 3(c) and the operational levers interact. Watch the agencies, not the press releases.