US Open-Source AI Lab Arcee Says Chinese Models Are Not Inherently Dangerous

As Chinese open-weight AI models grow in capability and popularity among U.S. companies, the debate over how to handle them has reached a fever pitch. The Trump administration has reportedly discussed banning Chinese AI models, while proprietary model makers like OpenAI and Anthropic have grown increasingly concerned about the competitive threat posed by low-cost Chinese alternatives like Moonshot AI's Kimi K3 and Alibaba's Qwen.

Into this charged atmosphere steps Arcee, a U.S. open-source AI lab that builds models for American enterprises. Its CTO, Lucas Atkins, told TechCrunch that Chinese open models are not inherently dangerous — a position that runs counter to the escalating push for restrictions.

"A lot of people view this as similar to a Chinese software program. Like, it was coded with these x, y, z intentions that a bad actor could simply command," Atkins said. "That is fundamentally not how these models are trained. There is really not any way for an Arcee, or an Alibaba, to make a model, have someone run it in their own environment and for us to have any access to it whatsoever."

Arcee would be among the most direct commercial beneficiaries of a Chinese model ban, yet Atkins argued that competition — not prohibition — is the right response. "I think instead of the conversation being about how to ban Chinese models, it should be about how to foster a good, open ecosystem here in the U.S.," he said.

Atkins noted that enterprises already subject open-weight models to rigorous security testing and post-training optimization before deployment, and that the risk of intentional backdoors in model outputs remains largely theoretical. "I don't know how you would do this," he admitted when asked about the possibility of models injecting malicious code.

He also acknowledged that Arcee itself learns from Chinese models. "We benefit from those models being good because we can learn what they did. We can build on top of them. Then they can learn what we do," Atkins said, expressing respect for the researchers building those models.

The interview comes at a pivotal moment for U.S.-China AI policy, with lawmakers divided between national security concerns and the practical reality that many U.S. companies rely on Chinese open models for cost-effective inference. Arcee's public stance — from a company that would profit from restrictions — provides a counterweight to the narrative that Chinese models pose a unique and unacceptable security risk.