Yc's tan advocates legal distillation over banning in AI debate

In September 2026, Anthropic released its second threat intelligence report, accusing Chinese labs of engaging in 'illegal distillation' by using stolen credentials to extract capabilities from cutting-edge models. CEO Dario Amodei had previously called for U.S. regulatory action against distillation. However, Y Combinator president Garry Tan argued against bans, suggesting instead that U.S. open-source labs should legally distill from advanced models through authorized channels. Tan emphasized that controlling API outputs from closed-source models is overly restrictive and that open-source weights should be treated as a public good. He also highlighted that proprietary labs, including Anthropic, used copyrighted materials during training, as seen in a $1.5 billion copyright settlement reached in July 2026. Tan warned against monopolistic dominance in AI, advocating for open-source models to drive progress and innovation. His stance came amid U.S. government discussions on AI distillation regulation, with YC representing the open-source startup ecosystem. Tan’s position, however, left unresolved key questions about the risks of open distillation and how to manage it effectively.