TAN PUSHES U.S. TO LET OPEN-WEIGHT AI LABS DISTILL FRONTIER MODELS

Tan's proposal centers on the idea that distillation—the process of extracting knowledge from larger models into smaller, more efficient ones—should be permitted for open-weight AI research in the United States. The Y Combinator president contends that frontier models built on public data create an obligation to democratize AI access. By allowing open-weight labs to distill these advanced models, the approach could accelerate innovation across the industry while reducing barriers to entry for smaller organizations. The stance reflects ongoing debates about AI model access, competition, and the role of frontier AI companies in supporting broader ecosystem development. It also highlights tensions between proprietary model developers and advocates for more distributed AI research capabilities. Tan's position suggests Y Combinator sees distillation as a pathway to balance commercial interests with public benefit in the rapidly evolving AI landscape.