White House officials are expected to revise the administration’s AI framework to cover powerful open models as well as closed frontier systems, according to WIRED.
The current framework reportedly focuses on the most capable closed models from major labs and would involve federal safety testing before public release. WIRED says an official indicated that open models could be added once they reach similar frontier capabilities.
The distinction matters because open models can be downloaded, modified, and deployed outside a provider’s direct control. That makes them valuable for research and competition, but also harder to govern with the same tools used for hosted commercial systems.
The framework has not been made public, so important details remain unclear, including thresholds, testing methods, and enforcement. The practical signal is that US AI policy is moving toward capability-based oversight rather than regulating only a model’s distribution style or the company that released it.