Lawmakers intensified scrutiny of DoorDash’s use of Moonshot AI’s Kimi model, demanding security records and executive briefings as Congress examines whether U.S. companies are leaning on cheaper Chinese open-weight systems.
Two House committees are escalating a joint investigation into U.S. companies using AI models developed by Chinese firms, with DoorDash facing a new demand for detailed disclosures over its use of Moonshot AI’s Kimi model. House Select Committee on the Chinese Communist Party Chairman John Moolenaar and Homeland Security Committee Chairman Andrew Garbarino asked DoorDash to provide by August 14 a full list of Chinese AI models it has used and all security testing records, and to make employees involved in AI infrastructure, software security, model evaluation, procurement, and legal or compliance review available for in-person briefings by August 21. The move follows public comments from DoorDash co-founder Andy Fang that the company had integrated Kimi into internal code review for lower-level software development work because it was cheaper and outperformed an all-U.S.-model setup, while harder tasks were reserved for Anthropic models. Lawmakers said the cost and performance advantages of Chinese open-weight models do not reduce national security concerns tied to dependence on systems developed under the jurisdiction of the People’s Republic of China. The broader probe also includes Coinbase, Airbnb and Siemens, reflecting growing congressional concern that U.S. companies are adopting models from Moonshot AI, DeepSeek and Alibaba’s Qwen for lower cost, customization flexibility and competitive performance. The investigation is also testing whether the United States has an adequate open-weight AI strategy as companies weigh expensive or restricted American models against cheaper Chinese alternatives.