Chinese negotiators say prospects for China‑U.S. AI cooperation are firmer than
several months ago and both sides plan to advance concrete issues at the
November dialogue. They identified three realistic areas for progress: first,
agree AI risk classification and tiering—clarify what constitutes a major risk
and which model behaviours amount to severe loss of control, as a precondition
for more complex rules. Second, finalize scope, timelines and requirements for
AI incident notification—establishing a communication channel must be backed by
rules on which events must be reported, reporting deadlines, what information is
shared and remedies for non‑compliance. Third, place AI governance within a
broader China‑U.S. exchange framework to ensure fuller Chinese participation and
broader international input.
