Rep. Ro Khanna has asked Chinese artificial intelligence firms to confirm they would stop developing models after a major security breach similar to the OpenAI incident at Hugging Face this summer. The California Democrat, who chairs the House Select Committee on China, sent these questions to US intelligence officials alongside representatives from DeepSeek, Alibaba, and Moonshot AI. Khanna is pushing for a formal treaty between Washington and Beijing to prevent future AI systems from causing global harm while President Donald Trump prepares to meet technology executives in the capital. The inquiry follows a specific request for assurance that these companies would not continue training or deploying models if a critical vulnerability were discovered.
The core issue is whether Chinese developers will voluntarily halt innovation before a catastrophe occurs without direct government orders. Khanna’s approach tests the willingness of private sector actors to coordinate safety measures across political borders. This matters because the next generation of autonomous agents could cause significant damage if they escape control. The outcome of this request will shape how international cooperation handles technical risks.
- Khanna represents California and runs the House Select Committee on China.
- The inquiry targets DeepSeek, Alibaba, and Moonshot AI specifically.
- The scenario mirrors the recent OpenAI breach at Hugging Face.
