Chinese AI Use Faces Washington Scrutiny

US lawmakers are scrutinising American companies’ use of Chinese artificial intelligence models, widening Washington’s technology concerns beyond advanced chips and infrastructure. The focus on DoorDash shows how the US-China AI rivalry is increasingly reaching routine corporate decisions over which models businesses deploy.

Lawmakers have questioned DoorDash over its reported use of Chinese-developed AI models, including technology from Alibaba. Their concerns centre on whether using such systems could expose sensitive information, strengthen Chinese AI companies or create dependencies on technology developed by a strategic competitor.

For DoorDash, the decision reflects the practical economics of deploying AI. Companies do not require the most advanced model for every task. Less expensive systems can be sufficient for routine workloads, allowing businesses to reserve more powerful and costly models for complex applications. As AI usage expands, that difference can materially affect technology spending.

This creates a challenge for US policymakers. Washington has used semiconductor export controls and other restrictions to limit China’s access to technologies needed to develop advanced AI. Software models are harder to contain, particularly when they are competitively priced, widely available and capable enough for commercial applications.

The debate also raises questions for corporate technology teams. Selecting an AI provider is increasingly becoming more than a decision about performance and cost. Data security, regulatory exposure and geopolitical risk are becoming part of procurement decisions, particularly when models are integrated into important business systems.

DoorDash therefore represents a broader issue facing American companies. Chinese developers do not necessarily need to outperform leading US models across every benchmark to gain adoption. They need to offer sufficient capability at an attractive price.

That changes the nature of AI competition. US leadership will increasingly depend not only on producing the most powerful models, but also on ensuring American technology remains commercially competitive for the everyday workloads businesses actually need.