Garry Tan, CEO of Y Combinator, is urging American open-weight AI labs to adopt distillation techniques for frontier models, arguing regulators should not intervene.
“I would do nothing,” he told CNBC earlier this week. “We could argue that there should be an American distillation regime.”
Tan explained to TechCrunch that he wants US-based open-weight labs to apply the same training methods used on American frontier AI labs. This approach would expand the selection of open-weight options available in the US, specifically those that are not Chinese.
Distillation involves a model maker prompting another model extensively to learn its reasoning and operation. AI labs commonly employ this legitimate technique to train new systems.
This week, Anthropic released a second report claiming Chinese labs are running “illicit distillation attacks.” These groups hide their identities to distill without permission, often relying on fraud and stolen credentials. Anthropic CEO Dario Amodei has previously asked US regulators to crack down on the practice.
It is notable that the head of Silicon Valley’s prestigious startup accelerator disagrees with that stance.
Tan is not advocating for American labs to use stolen credentials. He wants them to enter through the front door. His argument rests on two points. First, he believes it is an overreach for AI labs to dictate what their customers can do with information their models share. Second, he notes that proprietary labs did not ask permission when they ingested vast amounts of human knowledge to train their systems. They famously consumed copyrighted material without the consent of intellectual property holders.
“Controlling what users and customers do with API calls to closed weight models feels constraining, and there’s a role government can play here to normalize the fact that access to intelligence that was trained on broad public access data should itself also be more a form of a public good than something locked away behind restrictive terms of service,” he told TechCrunch when asked why American labs should be free to distill.
Tan, who once described himself as having cyber psychosis due to his heavy AI usage, wants to see a balance between open-weight AI labs and frontier labs.
“They are at the frontier and driving it forward. We want that to be fundable, and be a great business model ongoing,” he told CNBC. “You want open weight models to give people freedom and access.”
To him, the true doomer scenario for AI is for all the immense power of frontier AI to end up in the hands of a single powerful, proprietary provider.
“The nightmare scenario, the doomer scenario for AI is that there’s just one company,” he said. “It has the best access to capital. It has the best AI researchers. It runs away with it and suddenly there’s one company that’s monolithic. And that would be bad.”
What it means
For creators and developers, this stance suggests a future where access to advanced intelligence remains broad rather than consolidated. Tan argues that restricting how users interact with model outputs or preventing them from training new systems on existing ones creates unnecessary friction. If US labs adopt distillation openly, the ecosystem could retain more diversity, ensuring smaller players can compete with the large proprietary giants without facing legal barriers or regulatory crackdowns.
