Y Combinator CEO Garry Tan has publicly broken with the emerging consensus among frontier AI labs that distillation should be restricted. In interviews with CNBC and TechCrunch this week, Tan argued that US open-weight AI labs should be free to distill frontier models, directly countering Anthropic push for regulatory crackdowns.
I would do nothing, Tan told CNBC. We could argue that there should be an American distillation regime.
He wants smaller American open-weight AI labs to use distillation techniques on US frontier labs, giving the country a more robust set of open-weight options that are not Chinese. The statement comes days after Anthropic released its second report alleging that Chinese labs are engaged in illicit distillation attacks.
Tan argument has two prongs. First, he contends it is an overreach for AI labs to dictate what their customers can do with the information their models share via API calls. Second, he notes the irony that proprietary AI labs ingested vast amounts of copyrighted human knowledge without permission to train their own models, yet now seek to control what downstream users learn from those models.
The comments reveal a significant fracture in the AI industry response to the distillation debate, with the open-weight camp now openly advocating for the very practice that frontier labs are trying to shut down.
Tan also articulated what he sees as the real AI risk: not distillation itself, but concentration. The nightmare scenario for AI is that there is just one company, he said. It has the best access to capital and the best AI researchers. That would be bad.
