Meta Platforms Inc. CEO Mark Zuckerberg stated on July 28 that the US government should not ban Chinese AI models to gain advantage in the AI race. He warned that if leading US AI labs influence policy through 'regulatory capture,' it could stifle domestic market competition. Zuckerberg's comments come as Trump administration officials have presented evidence that Chinese AI startup Moonshot allegedly used 'distillation' technology to secretly train its models using US competitors' systems.
Zuckerberg Opposes China AI Model Ban
In an interview reported by the Financial Times on July 28, Zuckerberg said banning China's most advanced AI models from the US market is not an 'effective solution.' Multiple Trump administration officials stated they have evidence showing Chinese AI startup Moonshot used distillation technology to secretly and extensively leverage US competitors' models to train its own AI systems.
US Treasury Secretary Scott Bessent said last week that sanctions are not ruled out for Chinese AI labs that 'cross the line in stealing intellectual property.' However, Zuckerberg opposes implementing such bans, stating that US companies should systematically identify development bottlenecks and obstacles to enhance their ability to compete with China.
Meta CEO Advocates for Open-Source AI Models
Zuckerberg wrote in the Wall Street Journal on July 28 reiterating his support for 'open models.' He warned of the risks if AI becomes concentrated in the hands of a few large corporations. Outside observers noted these remarks clearly reference Anthropic and OpenAI, two companies currently dominating the chatbot market with proprietary 'closed' AI models.
Zuckerberg also discussed how the US should manage the release of next-generation AI models. He stated that whether using peer review or vetting mechanisms, as long as they are 'properly executed by people with professional judgment,' they can help the AI industry develop. However, he warned against allowing leading AI companies to have excessive influence over related regulatory mechanisms.
Analyst Highlights AI Market Concentration Risks
Gavin Baker, managing partner of hedge fund Atreides Management and early SpaceX investor, stated that Moonshot's new-generation model 'Kimi K3' could be an important turning point for AI, with potential negative implications for Anthropic and OpenAI, but representing a net positive for other sectors including power, semiconductors, hyperscalers, neoclouds, and software.
Baker wrote on social platform X on July 17 that if only 2-3 frontier labs dominate the market and capture 90% of AI inference profits, this would bring net negative effects to all other AI industry chain participants. These labs would become monopsony buyers for power, data centers, semiconductors, and hyperscale cloud services, and over time would vertically integrate to control computing power and hardware, while absorbing the application layer and software companies.
Baker stated directly that any factor that squeezes model layer profit margins and promotes competition would transfer profit margins to semiconductors, power, hyperscalers, neoclouds, and software application layers. This is also why Nvidia Corp. CEO Jensen Huang strongly supports open-source models.
FAQ
What position did Zuckerberg take on banning Chinese AI models on July 28?
Zuckerberg stated in an interview reported by the Financial Times on July 28 that banning China's most advanced AI models from the US market is not an 'effective solution.' He believes US companies should systematically identify development bottlenecks to enhance their ability to compete with China, rather than implementing bans.
Why does Zuckerberg warn against regulatory capture in AI?
Zuckerberg warned on July 28 that if leading US AI labs have excessive influence over regulatory mechanisms, it could stifle domestic market competition. He stated that regulatory mechanisms should be executed by people with professional judgment, but leading AI companies should not have oversized influence over related policies.
