White House will exempt 'open' AI systems from security review, report says

The White House will exempt “open” AI systems from security review, according to a report from The Washington Post. Systems classified as “open” would not be subject to the security review process applied to AI models, separating their regulatory path from that of closed systems.

The move is the latest turn in the U.S. debate over open versus closed AI. For years, policymakers have argued about the security risks of open-weight models: one camp says openness fuels innovation and competition, while the other warns that such models can be misused and should face stricter scrutiny and controls.

For open-source developers, an exemption would mean a lower compliance burden. Uncertainty over security reviews has been seen as a potential drag on open-source AI development, and the White House's reported approach would leave the open-source route on a lighter regulatory footing.

The direction is likely to be contested. Security researchers have repeatedly warned that open-weight models exempted from review could leave room for misuse, and the definition of what counts as “open” is itself expected to become a point of contention among lawmakers and regulators.

The report does not detail the exact definition of “open” systems, the scope of the review regime, or a timeline for implementation. What to watch next: the wording of official White House documents, which systems qualify as “open,” and how the policy shapes the U.S. AI industry and the broader competitive landscape.