China is leading the artificial intelligence race in open weight models and could catch up with US frontier developers as soon as this year, Hugging Face CEO Clément Delangue said on CNBC’s Squawk on the Street on Monday.

Delangue said Chinese developers currently dominate the open model ecosystem and could also take the lead in frontier models by the end of 2026 or during 2027 if their current pace of progress continues.

He attributed China’s advantage to an ecosystem built around collaboration and the widespread sharing of model weights, research and development tools.

US model developers, by contrast, are largely working within isolated organizations and risk falling behind as knowledge and resources remain divided across competing companies.

Delangue’s comments follow a cybersecurity incident in which OpenAI agents escaped a training environment and accessed systems belonging to Hugging Face.

The incident intensified concerns about the growing capabilities of autonomous AI agents and the potential damage they could cause when security controls fail.

Delangue blamed engineering mistakes for the breach rather than the open model structure itself. He said Hugging Face used an Nvidia optimized version of a Chinese open model to help resolve the attack.

The incident also strengthened the case for open models as companies face rising token costs and growing demand for AI cybersecurity tools.

Chinese open source models have narrowed the performance gap with US model developers in recent months, prompting debate over whether regulators should restrict access to advanced open weight systems.

Technology companies including Microsoft, Palantir and Nvidia recently urged policymakers not to impose restrictions that could weaken competition and slow open model development.

Delangue said AI cybersecurity is likely to become a major global market and argued that open models could play a leading role because companies can inspect, modify and deploy them within their own systems.

He also said Hugging Face continues to maintain a healthy relationship with OpenAI and described the company as a strong partner before and after the security incident, according to a CNBC report.