US lawmakers investigate DoorDash’s use of Moonshot AI’s Kimi K2.6 model

In the rapidly evolving landscape of artificial intelligence, China has made significant strides, closing the gap with the United States in just a few months. While US giants like Anthropic and OpenAI have poured billions into cutting-edge AI models, Chinese firms have adopted a more open-source approach. Companies like DeepSeek, Alibaba's Qwen, and Xiaomi's MiMo have developed dozens of high-quality AI models that are now widely used in businesses, government services, and consumer devices.

A recent development is Moonshot AI's unveiling of Kimi K3, a model that can rival Anthropic's Claude Fable 5, even matching or surpassing it in some key benchmarks. Analysts suggest this could make high-end AI more accessible and affordable, making it a "commodity" for most customers. However, this progress has not gone unnoticed by Washington.

US President Trump has threatened to block access to Chinese AI models, and Treasury Secretary Scott Bessent has warned of potential sanctions over alleged intellectual property theft. US AI leaders, Anthropic and OpenAI, accuse Chinese labs of distilling their models, a practice of using AI systems to train others on a large scale.

China has rejected these accusations as baseless, calling US threats "technological bullying." While China's success in AI may be partially attributed to talent, infrastructure, and resources, analysts like Giulia Neaher argue that open-source models have played a significant role. Open-source AI allows for global collaboration and fine-tuning to suit specific needs, making it easier for Chinese labs to harness the global AI community's knowledge.

Despite China's current advances in pricing and deployment, the US remains committed to maintaining its lead in frontier AI, as control of advanced systems could provide military, economic, and technological dominance for years to come.