Leading tech companies including Nvidia and Palantir are urging the US not to restrict "open-weight" AI models, despite concerns over intellectual property theft by Chinese labs. Open models allow users to customize and run AI on their own hardware, gaining popularity among Chinese developers who have shown recent technical advancements. This plea comes as Washington considers targeting Chinese AI firms, particularly after the release of Moonshot AI's Kimi K3, which reportedly used stolen technology from Anthropic.

US officials have expressed strong concerns about "covert industrial distillation aimed at stealing proprietary US technology." While sanctions are being considered, there's also support for open-weight AI. The debate intensifies as companies like OpenAI and Anthropic, which produce closed models, aim for multi-trillion dollar valuations by maintaining a lead over foreign rivals.

Nvidia CEO Jensen Huang advocates for embracing open-source models, including Chinese ones, believing they drive innovation and usage. He also supports selling AI chips to China to maintain US industry leadership. The companies advocating for open models acknowledge inherent risks but argue that open examination by researchers can lead to better safeguards and improvements over time, rather than increased risks from suppression.