As Chinese open-weight AI models grow in capability and popularity, arguments over what should be done about them have once again reached a fever pitch.

There is talk that the Trump administration could try to ban them (although it has not yet acted on the idea). Meanwhile, proprietary model makers, particularly OpenAI and Anthropic, seem increasingly concerned about them.

Open-source models like Moonshot AI’s Kimi K3 or Alibaba’s Qwen offer inference at a fraction of the nominal cost of closed-source models from these big American labs. The fear is that they also represent some kind of threat. They certainly threaten the profit margins of the large proprietary AI labs.

But should companies running these models in their own data centers succumb to fear that they could be a vector for Chinese hackers?

No, says Lucas Atkins, chief technology officer at Arcee, which is building open models to give American companies a local alternative to Chinese models.

If any startup would benefit from banning Chinese models, Arcee would. But Atkins says China’s open models are no more dangerous than any other open source software a company might use. In fact, he says, they even offer benefits to his own company.

“A lot of people see this as something similar to a Chinese software program. It was coded with these x, y, z intentions” that a bad actor could simply command, he said.

“Basically, that’s not how these models are trained. There’s really no way for Arcee or Alibaba to create a model, for someone to run it in their own environment, and for us to have access to it,” he explained.

While most of these models are what is known as “open weight” and are not actually open source software, the source code (the part that will actually run on the servers), if downloaded from open source sites like Hugging Face, is still viewable and largely reviewable. (What is not available are the methods and data used to train the models.)

Large organizations must subject any core models to their security inspection and testing processes, and will often also subsequently train the models for their specific uses and may examine areas such as bias, toxicity, hallucinations, and sensitivity to certain topics. So they work, optimize, and understand the models before people start sending them directions.

Could a model that is used for coding somehow introduce malicious backdoors into the code it writes? Again, while that is theoretically possible, it would take acrobatic feats to achieve.

“There’s no reason why a sufficiently sophisticated actor couldn’t train a model to be a completely amazing coding model under all circumstances, but when presented with a certain type of code base… some hidden training would be triggered,” postulated Atkins, who spends his days training models. But he adds: “I don’t know how you would do this.”

Because large language models are creative by nature, the odds of getting a contemporary model to spit out malware in response to a preplanned perfect storm of context and warning are slim. Even slimmer are the chances of any company using that code.

Could it happen in the future? That’s a mystery. But companies are also building their AI applications to be model-agnostic and use multiple models. So even if the Chinese models are the best for the current price, companies won’t be forced to use them forever.

“I think instead of talking about how to ban Chinese models, we should focus on how to foster a good, open ecosystem here in the United States,” Atkins says.

Arcee also gains advantages from Chinese models. Because they are open, the startup “benefits from those models being good because we can learn what they did. We can build on them. Then they can learn what we do,” he says. “We have enormous respect for the people who build those models, the individual researchers.”

Ultimately, the way to compete with Chinese models “is to launch a model that is better,” Atkins says. “We need to give them something to talk about.”