White House New AI Rules Exclude Open-Weight Models from Safety Testing

DavidSKrueger · x · 2026-08-06

Under the new AI regulations introduced by the White House, open-weight models will not be safety-tested as part of the voluntary program.

This policy shift has sparked concerns within the AI safety and open-source communities. Critics are questioning whether the White House is discriminating against open-weight models, fearing it could hinder the open-source ecosystem and alter the current landscape of model safety evaluations.

Related event: White House AI Rules: Open-Source Exempt, Closed-Source 30-Day Review(12 posts)→

Original post →

More from Safety

Safety channel →