Hugging Face CEO says China is winning the AI race while the US is building 'in silos'

Aug 04, 2026 05:50 AM - 1 hour ago 3

Clément Delangue, CEO of Hugging Face, arrives for the Inaugural AI Insight Forum successful Russell Building connected Capitol Hill connected Wednesday, September 13, 2023. (Tom Williams/CQ-Roll Call, Inc via Getty Images)

The Hugging Face CEO is making a lawsuit for open-weight AI models — pinch a informing for the US.

Clément Delangue said connected CNBC's "Squawk connected the Street" show connected Monday that China is winning the AI title because the state is "more unfastened subject and unfastened models than the US."

"I deliberation they are," Delangue said Monday connected CNBC erstwhile asked by big Sara Eisen whether China is winning the AI race. "So there's this benignant of emulation and the complaint of advancement is overmuch faster location than successful the US, wherever everyone is building successful silos successful immoderate of the frontier labs and not sharing pinch the remainder of the ecosystem."

He added, "I wouldn't beryllium amazed if they commencement dominating astatine the frontier successful general, not conscionable unfastened models, either by the extremity of this twelvemonth aliases adjacent twelvemonth astatine the complaint of progress."

The statement complete open-weight AI has intensified aft a July incident successful which an OpenAI agent escaped a training situation and hacked Hugging Face's platform. Delangue antecedently said he had turned to GLM 5.2, an open-source exemplary from Beijing-based Z.ai, to thief take sides the company's system.

"We were attacked by an unreleased backstage exemplary built down closed door," he said. "And we could only take sides ourselves pinch unfastened models because the guardrails of the APIs didn't fto us."

"I deliberation successful the early we'll spot that astir of the attacks are really going to travel from backstage proprietary models because the attackers, they don't attraction astir guardrails," he added. "They don't attraction astir position of service, and astir of the defense will really beryllium done by unfastened models."

The remarks onshore arsenic policymakers statement whether open-weight AI from China should look tighter restrictions.

In precocious July, complete 2 twelve tech companies, including Nvidia, Microsoft, Meta, and OpenAI, signed a letter urging US policymakers not to ace down connected open-weight AI models. Days later, Anthropic CEO Dario Amodei clarified his company's stance connected the rumor in a blog post. "Open-weights models that don't person vulnerable capabilities are a nationalist good," he wrote.