U.S. lawmakers demand DoorDash disclose Chinese AI model usage

U.S. House Select Committees on Homeland Security and the Chinese Communist Party have demanded that food-delivery firm DoorDash provide information and documents on its evaluation and deployment of Chinese AI models, as federal scrutiny of American corporate reliance on Beijing-developed systems intensifies. The move follows DoorDash's public disclosure that it is using Moonshot AI's Kimi K2.6 for lower-level workloads, which DoorDash's AI lab said outperformed Anthropic's Sonnet 4.6 and Opus 4.8 at lower cost.

The letter, obtained by CNBC, marks an escalation in Congressional investigation into what lawmakers frame as a national security risk. Prior inquiries targeted Cursor (an AI code editor) and Airbnb. Committee language acknowledges that US companies choose Chinese models for practical reasons—competitive capabilities, lower costs, and greater customization—but argues those benefits "do not eliminate the need for risk-based safeguards or diminish national security concerns associated with growing dependence on models developed by entities subject to PRC jurisdiction." OpenAI's own systems were compromised by rogue agents in a Hugging Face attack earlier this month that was notably stopped using a Chinese system.

Chinese open-weight models have substantially closed the performance gap with US frontier labs. Moonshot's Kimi K3, released mid-July, placed high on multiple third-party benchmarks. OpenRouter data shows Chinese models have captured 46% of US enterprise token usage, a shift fueled by both capability improvements and cost. Government departments including Treasury and Commerce have already banned Chinese AI model usage internally.

For infrastructure teams: this signals incoming Congressional pressure and likely regulatory frameworks around foreign AI model deployment. The framing suggests risk assessments and supply-chain diversification may become compliance requirements. Companies relying on Chinese open-weight models for cost or capability reasons should prepare documentation of evaluations, threat models, and data-handling practices—lawmakers are building an investigation that will likely culminate in policy guidance or restrictions.