Huge debate in Silicon Valley is growing due to the proliferation of Chinese-made artificial intelligence tools, particularly “open scale” artificial intelligence systems that in some respects can rival or even outperform some of the best American models. My WIRED colleague Hugo Lowell wrote about the Trump administration’s internal debate about how to deal with Chinese models. Among artificial intelligence companies in the Valley, the issue is even more divisive.
The biggest problem in both DC and the Valley is distillation, where a less powerful AI model is trained on the results of a more powerful model. In June, Anthropic accused Chinese tech giant Alibaba of illegally stealing its IP address through distillation attacks. Then, earlier this week, the White House said it believed Beijing-based Moonshot AI developed the Kimi K3 model by distilling Anthropic’s Fable 5 model.
Another major problem is how quickly Chinese models appear and spread. The core components of an open-weighted AI model are made publicly available so that they can be customized to meet user needs. However, they do not have the same guardrails on which Anthropic builds its reputation. Yasir Atalan, deputy director and data specialist at the Center for International and Strategic Studies, points out that the main advantage of open AI models is their speed diffusion. They can spread particularly easily “through Hugging Face, GitHub, cloud service providers, on-premises deployments, and third-party inference platforms” I’m writing. If you’re an anthropologist and have built a cult around security and access fees for your immense, costly, proprietary models, you have every reason to want to regulate it.
But some Silicon Valley startups — not the trillion-dollar ones like OpenAI and Anthropic — don’t actually want the U.S. government to place limits on these AI models. On Wednesday, a group of over 200 startups named the Little Tech Association sent a letter Michael Kratsios, President Donald Trump’s science adviser, and U.S. Secretary of Commerce Howard Lutnick lobbying against a blanket ban on open-weight AI models. The group, which includes famed startup incubator YCombinator, favors some protections but argues that denying Americans access to AI models abroad would weaken American startups and create a monopoly among AI giants.
Bill Gurley, legendary technology investor and longtime partner at Benchmark Capital, has publicly advocated for allowing “free market work” In long blog which offers an interesting history of open source software, Gurley writes that open-source models avoid closure, encourage real academic research, and are crucial for capital-constrained startups.
“Every AI startup, every individual developer, every two-person team building a product on AI infrastructure depends on access to good models at affordable prices,” Gurley says.
Chamath Palihapitiya, one of Everything in podcast hosts, wrote on X that “fooling the U.S. government into protecting the border lab business model with a China bogeyman is a mistake… It protects the capital of the 5,000 people who invest in OAI and Ant while selling everyone else. That would be a terribly stupid decision.” His co-host and fellow VC Jason Calacanis spared no effort. “Dad Trump, protect us!!!!” He wrote on Xwith an alarming number of crying and laughing emojis.
This stance by some of Silicon Valley’s most ruthless capitalists may seem counterintuitive at first. Why allow a foreign adversary’s technology to flourish in the US? It’s as if the United States was only 1-0 up in the AI World Cup, a somewhat uncomfortable lead, and the fans were chanting for the other team to get a free kick.