Kimi K3 Is So Powerful OpenAI's Strategy Lead Is Worried About US Defenses

Former White House AI advisor and OpenAI strategy lead Dean W. Ball says Kimi K3 is a seriously strong model. Its performance in agentic programming is already approaching the best publicly available models from early 2026. That kind of capability can't be explained by distillation alone.

But Ball isn't just praising Kimi K3.

What he really wants to discuss is why China would release a model of this caliber as open source — and how it could reshape the US AI industry.

In his view, the pressure from Chinese open-source models isn't just about having a cheaper competitor. If open models are strong enough, developers won't keep paying high prices for closed-source models. Margins for model makers shrink, investors get cautious. The incentive for US companies to spend hundreds of billions training next-generation models could fade.

Open weights let technology spread faster, but they also make it harder to profit from building frontier models. Eventually, model development might have to be subsidized by other businesses or funded by the government, turning AI into a public utility like the power grid or highways.

Ball speculates that China is willing to do this partly because it doesn't take advanced AI risks seriously enough. Another reason: after US restrictions on advanced chip exports, China lacks the compute capacity to offer inference services globally. If it can't lock users in through APIs, releasing open weights becomes a way to expand influence.

He also predicts the US government will eventually try to keep Chinese models out of American enterprises. The US doesn't need to ban open-source models outright — just continually raise red flags about backdoors, data security, and compliance risks. Regulated industries like banking will steer clear on their own.

Ball even thinks these warnings don't need particularly strong evidence. Just create enough uncertainty to make companies afraid to adopt the models, without pushing all developers toward harder-to-regulate overseas providers.