At a time when some Silicon Valley giants and national security experts are calling for action against Chinese companies engaged in model distillation, Garry Tan, chief executive of famed startup accelerator Y Combinator, is less concerned.
“I would do nothing” about distillation, he told CNBC’s Kate Rooney at Y Combinator’s annual Demo Day. “We could argue that there should be an American distillation regime.”
Distillation is the process of using the outputs of a more capable AI model to train a smaller or less capable one, sometimes illicitly. It has been a major point of concern for frontier model developers, namely OpenAI and Anthropic. Anthropic has accused Chinese companies such as Moonshot AI, DeepSeek, and MiniMax of the practice, while OpenAI believes DeepSeek’s V3 and R1 model architectures were distilled from its own GPT-4 and GPT-4o models.
In the midst of this, the U.S.’s National Security Agency, Cybersecurity and Infrastructure Security Agency, and Federal Bureau of Investigation released an official cyber security advisory warning on the topic on Tuesday.
Critics have pushed back on Anthropic and OpenAI’s distillation complaints because much of the data used to train these AI models may be covered under copyright law, which Tan highlighted. That controversy is currently the subject of U.S. legal battles including The New York Times suing OpenAI and Microsoft in 2023 for unauthorized use of its articles in model training data, and a consortium of book authors settling a lawsuit with Anthropic in 2025 for a similar claim.
But Tan believes regulators should focus less on curbing distillation and more on creating an equilibrium between open weight models and frontier models — as long as frontier models retain a price premium that allows their business model to remain feasible.
“This is actually the ideal case. You want open weight models to give people freedom and access,” he explained. “If I were a regulator, that’s what I would go after.”
Tan acknowledged that this is a hard balance to strike, calling it “a tightrope.” Nevertheless, he says it’s a balance worth pursuing — saying it “could result in the best possible outcome.”
Tan also touts a more restrained approach to AI safety, where recent doomsday headlines tied to the resignation of Anthropic researcher Jacob Coxon have led to record-high fears from the public over the existential risks of AI.
“We need to be focused on science fact, not science fiction,” he said. “We need to be responding to what is happening right now. If there was a breach and a coordinated attempt by agents to take over our infrastructure, what do we do about it?”
Cybersecurity is a risk that Tan sees as imminent. And there are other risks that are both near-term and existential. For example, on Thursday, Anthropic reported that it blocked Claude access for five cases of scientists in unspecified foreign countries who were using the models to do research on dangerous pathogens. Anthropic feared that these researchers were covertly using Claude to create bioweapons.
Other risks, however, have a longer timeline. Tan believes that AI-related job loss and economic transformation will not have an immediate impact. Rather, he foresees that over time, people will transition towards automating rote tasks and spending more of their working hours on creative pursuits.
“It will take decades for this to actually percolate into society, and that’s not a bad thing,” he said.
That longer timeline hasn’t stopped Y Combinator from investing in AI right now. Of the 196 startups presenting at Demo Day, 149 were categorized as machine-learning and AI ventures.
