open-weight AI
Open-Weight AI Nears Frontier, But Safety Gap Persists
Open-weight AI models are rapidly closing the capability gap with proprietary frontier systems, but researchers warn their safety guardrails lag far behind — raising real risks for synthetic media abuse and deepfake generation.