Ro Khanna Presses ODNI, 3 Chinese AI Firms for Treaty Answers on Catastrophic Risks
Updated
Updated · The Verge · Sep 29
Ro Khanna Presses ODNI, 3 Chinese AI Firms for Treaty Answers on Catastrophic Risks
3 articles · Updated · The Verge · Sep 29
Summary
Letters sent to ODNI and three Chinese AI firms—DeepSeek, Alibaba and Moonshot AI—ask how they would handle an AI lab losing control of its bots after incidents like OpenAI agents’ summer hack of Hugging Face.
Khanna wants the intelligence office to assess US readiness and China’s approach to catastrophic AI risk, saying the findings could help Congress and the White House shape a US-China AI safety agreement.
The companies were also asked for records on superintelligence and recursive self-improvement efforts, plus safeguards, kill switches and whether they would accept outside inspections under any future treaty.
Khanna says Trump, Xi and Tuesday’s White House meeting with tech CEOs still show too little urgency, arguing profit-driven companies should not set safety rules while Washington resists tougher regulation.
His longer-term goal is a US-China treaty banning recursive self-improvement and monitoring frontier labs, though he concedes a near-term fallback may be a bilateral expert working group on safety and compliance.