US Companies’ Shift to Chinese AI Models Triggers Joint Congressional Probe

Chinese open-source models such as DeepSeek and Z.ai’s GLM are approaching the performance of Anthropic and OpenAI while costing 60% to 90% less, reshaping AI procurement by US companies. The trend threatens revenue at US model providers and raises questions about corporate data, cybersecurity, and Washington’s policies for maintaining its AI lead.

In April 2026, the US House Committee on Homeland Security and the House Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party announced a joint investigation. On June 4, AI startup Lindy shifted 100% of its traffic from Claude to DeepSeek, estimating that the move could save several million dollars within months. OpenRouter data showed that Chinese models accounted for more than 30% of weekly token usage from February 8, peaking at 46%.