America and China do not agree on much.But when it comes to AI, we all lose if we lose control.Last month, as the ranking member of the House Select Committee on China, I convened the first-ever congressional hearing calling for a binding, enforceable, and verifiable U.S.-China AI treaty.
I also led every Democrat on the Committee in sending the White House a letter demanding global action to tackle AI risks.Within days, President Trump and General Secretary Xi announced that they would establish a bilateral dialogue on AI.
This is an important first step that would not have happened without congressional pressure.But this moment demands more than discussion.At my hearing, some of the world's leading experts on AI called for the U.S.
and China to form joint technical working groups with both countries' top scientists.Their mission would be to develop mutually trusted verification technologies that prevent either side from cheating once we sign a U.S.-China AI treaty.
This is an easy second step we should take today.ANTHROPIC CEO LIKENS AI FIGHT WITH CHINA TO COLD WAR, SEEKS 'DISARMAMENT NEGOTIATIONS'Critically, we also need to begin envisioning what an actual treaty to save humanity from unchecked AI would look like.The risk that humans will lose control of AI or that misaligned models could unleash biological havoc is increasing every day, and if the world's AI superpowers do not act soon, it may be too late.To address these risks, there are five key pillars of a U.S.-China AI treaty.
First, we need a worldwide pause on recursive self-improvement.This is when AI develops AI without human intervention, and it's the most likely path to humanity losing control of this technology entirely.
Asking AI to design its own safeguards is like asking the fox to guard the henhouse.MIKE JOHNSON REJECTS AI MORATORIUM OVER CHINA THREAT AS CONSERVATIVE ACTIVIST DEMANDS GUARDRAILSSecond, we should impose a global pause on superintelligence.As it has been traditionally defined...