Will Chinese AI companies slow down? A top House Democrat wants answers

Rep. Ro Khanna, the top Democrat on the House Select Committee on China, is pushing a simple idea with very large consequences: get the US and China to agree on rules that stop AI from causing chaos. He’s also asking whether anyone is actually ready for the kind of incident that could expose how fragile these systems are. Not just in theory. In practice.

The timing is not subtle. President Donald Trump is preparing to meet tech and AI executives in Washington, and Khanna is using the moment to call for a treaty between the two biggest powers in the room. The problem, as the article makes clear, is that this sort of agreement would be extremely hard to land. Both governments would have to move, and quickly. That’s not exactly their favorite hobby.

In letters shared exclusively with The Verge, Khanna asked a US intelligence agency and several Chinese AI companies if they would be prepared for an incident like the one this summer involving OpenAI agents and a hack of Hugging Face. The companies he named were DeepSeek, Alibaba, and Moonshot AI. The question is blunt because the risk is blunt: AI systems don’t need to be science-fiction smart to cause real-world messes.

Khanna is not treating this as a niche technical issue. He’s framing it as a geopolitical one, with the possibility that one bad incident could spill well beyond a single company or country. That’s the angle here: before the world gets another dramatic AI security lesson, he wants both sides to say what they’d actually do when the systems misbehave.