Hugging Face CEO Clément Delangue has asserted that China is currently winning the artificial intelligence race due to its more open approach to AI models and scientific collaboration. Speaking on CNBC, Delangue contrasted this with the United States' tendency to develop AI in 'silos' within frontier labs, hindering broader ecosystem progress.
Delangue expressed concern that China's rapid progress could lead to its dominance in frontier AI within the next year or two. His remarks come amid an intensifying debate over the merits and risks of open-weight AI models, particularly those developed in China.
This discussion has been fueled by recent security incidents. Delangue revealed that Hugging Face's platform was attacked by an unreleased private model from OpenAI, and the company could only defend itself effectively using open-source models. He argued that future attacks are more likely to originate from private, proprietary models, as attackers disregard guardrails and terms of service, making open models crucial for defense.
Policymakers are currently debating whether to impose stricter regulations on open-weight AI models from China. In response, over two dozen tech companies, including Nvidia, Microsoft, Meta, and OpenAI, signed a letter urging against such crackdowns. Anthropic CEO Dario Amodei also weighed in, stating that open-weight models without dangerous capabilities serve as a public good.
