Trump AI Framework Excludes Open Models
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Trump administration released a voluntary AI cybersecurity testing framework that explicitly excludes open models from review, allowing unrestricted post-release distribution
- Framework guidelines apply only to closed-source, state-of-the-art AI models deemed to carry national security risks, though the government does not define what qualifies as such
- AI companies including OpenAI, Anthropic, and Google attended a White House briefing on the finalized framework, which grants a 30-day grace period for federal review before model release
- Executive Order signed by Trump in June directed AI firms to share frontier models with the federal government ahead of launch to address cybersecurity concerns
- Open models are not only excluded from testing but are protected from future restrictions under the framework, which states it cannot be used to limit them after release
- Smaller AI providers face uncertainty due to the lack of definitions for 'frontier,' 'state-of-the-art,' or 'national security risk,' making compliance expectations unclear
Why it matters: Major AI labs gain regulatory clarity while smaller developers face uneven ground due to undefined thresholds, and the exclusion of open models removes oversight from a growing segment. The lack of clear criteria means national security judgments rest on subjective interpretation, increasing compliance risk for less-resourced actors.
Ask SkimNews
