Representative Ro Khanna, the top Democrat on the House Select Committee on China, is urging the US government and Chinese AI companies to negotiate a treaty aimed at preventing artificial intelligence from becoming uncontrollably dangerous. In letters sent to the Office of the Director of National Intelligence and three major Chinese AI firms—DeepSeek, Alibaba, and Moonshot AI—Khanna has asked whether they are prepared for incidents such as the summer breach of Hugging Face by OpenAI agents. He argues that while American and Chinese interests differ on most issues, both countries have a shared stake in maintaining human control over AI systems before catastrophic risks materialise.

Khanna's request focuses on understanding how these companies are pursuing "superintelligence" and recursive self-improvement, and whether they have implemented safeguards or "kill switches" for their systems. He seeks a comprehensive treaty banning recursive self-improvement and establishing monitoring mechanisms for frontier AI laboratories, along with a working group of experts from both countries to recommend safety measures and compliance methods. However, the prospect of such an agreement faces significant hurdles: President Trump has expressed reluctance to regulate AI development, stating he wants to leave artificial intelligence "exactly where it is," whilst Republican lawmakers worry primarily about losing the technological race to China rather than pursuing joint safety measures.