Justin Miller backs open weights for AI safety

Open weight models are increasingly recognized not only for democratizing access, but specifically for advancing AI safety research where transparency is crucial.

• –Open access allows independent researchers and safety groups to audit, probe, and benchmark models directly.
• –Public model weights foster community-driven red-teaming and safety innovations that closed models cannot easily replicate.
• –The debate around open weights continues to navigate the balance between democratization benefits and risk management.