White House Develops AI Framework Excluding Open Models

The White House is excluding open models from its framework to test advanced AI capabilities, according to sources familiar with the matter. The voluntary framework will define how the administration reviews advanced AI models before release, but the details are not being made public. The framework defines a covered frontier model as closed-source with state-of-the-art capabilities and national security risks, but there is no clear definition of what is considered state-of-the-art or a national security risk.

The framework explicitly excludes open models, and notes that nothing in it should be interpreted as restricting open models once they've been released. During the 30-day pre-release government review period, employees would be limited from accessing models, which would be stored in high-security environments with detailed logs on who is accessing the models. The review process will include various administration officials rather than a single office or agency.

The White House does not plan to publicly release the framework, and companies that weren't invited to staff-level meetings on Tuesday remain in the dark about its contents. It's also unclear which 'trusted partners' will get early access to advanced models under the framework, including whether any foreign governments would qualify. The executive order behind the framework explicitly says the benchmarking process to assess advanced cyber capabilities of AI models will be classified.