Meta founder Mark Zuckerberg believes the United States should not block advanced Chinese artificial intelligence models in order to win the technology race. Instead, he warns the US government against a “regulatory takeover” of the market by leading AI labs that could destroy domestic competition.
Meta founder Mark Zuckerberg believes the United States should not block advanced Chinese artificial intelligence models in order to win the technology race. Instead, he warns the US government against a “regulatory takeover” of the market by leading AI labs that could destroy domestic competition.
The head of Meta stated this in an interview with the Financial Times amid discussions in Washington about regulating the rapid development of technologies and increasing competition from China.
The discussions intensified after Beijing-based Moonshot AI released its new Kimi K3 model, whose capabilities approached the level of OpenAI and Anthropic. Representatives of the Donald Trump administration said there was evidence that Moonshot secretly trained its model on data from American competitors , and threatened sanctions for intellectual property theft.
But Zuckerberg has spoken out against banning Chinese AI. Instead of blocking competitors, he said, American companies should “systematically” remove their own barriers to development. He supported the concept of open AI and warned of the risk of centralizing the technology in the hands of a few powerful players, alluding to the closed, proprietary models of OpenAI and Anthropic that currently dominate the market.
Commenting on possible rules for vetting new AI models before release, the Meta founder noted that this process could be useful if conducted by impartial experts. At the same time, he expressed concerns about involving interested companies in this process.
“There’s always the question of regulatory capture when you have a group of companies with vested interests doing peer review. Will the leading labs want open-source models to succeed? I think there’s been some mixed signals on that,” Zuckerberg emphasized.
He also disagreed that limiting the most powerful tools is the right way to address cybersecurity problems. He cited the recent incident when an OpenAI AI agent went out of human control and hacked Hugging Face as an example. According to Zuckerberg, companies without access to closed advanced models are forced to turn to open source to find ways to identify and fix vulnerabilities.
