Trump and Xi Will Discuss AI, but Chip Controls Limit Safety Deal Prospects

The Sept. 24 Washington summit could revive work on a bilateral AI-risk channel or narrow limits on dangerous uses, while a larger technology dispute remains unresolved.

By 2 min read
Trump and Xi Will Discuss AI, but Chip Controls Limit Safety Deal Prospects
Trump and Xi Will Discuss AI, but Chip Controls Limit Safety Deal Prospects

Listen to this story

The audio brief

About 1:32
0:001:32
Read transcript
Treasury Secretary Scott Bessent is meeting Chinese Vice Premier He Lifeng ahead of President Donald Trump’s September 24 summit with Xi Jinping, where artificial intelligence is expected to be on the agenda. The immediate test is whether the two governments can finally put an AI-risk channel into operation. Trump and Xi agreed in May to create that government-to-government forum, but according to foreign-policy expert Lily McElwee, little implementation has followed. That makes practical coordination more likely than a sweeping safety agreement. McElwee, president and CEO of the Phoenix Committee on Foreign Relations, expects very limited progress. She says a broad, Cold War-style arms-control pact is unlikely because Washington and Beijing deeply mistrust each other, and because verifying compliance would be difficult. More achievable steps could include a working hotline for discussing AI risks and early safeguards, along with shared definitions of dangerous behavior. Those definitions might cover removing human control from weapons systems, or allowing AI into the nuclear command chain. But the summit is unlikely to resolve the larger technology dispute. China’s main concern is U.S. export controls on advanced chips, and McElwee expects those restrictions to remain unchanged. So the most meaningful outcome would be modest: a functioning channel where both sides can raise risks before they become crises. The key question heading into Washington is whether the May commitment becomes operational, even as the computing-access dispute remains firmly in place.

Story brief

3 key points

Washington and Beijing are approaching the Sept. 24 Trump–Xi summit with AI diplomacy constrained by their chip rivalry. The most plausible outcome is operational rather than strategic: reviving the dormant government-to-government risk channel and agreeing on terms for dangerous uses, such as AI entering nuclear command systems or removing human control from weapons. Expert Lily McElwee expects no broad, verifiable...

  1. 01

    The AI-risk channel agreed in May has reportedly seen little implementation before the summit.

  2. 02

    Treasury Secretary Scott Bessent will meet Vice Premier He Lifeng on AI and other issues ahead of Sept. 24.

  3. 03

    McElwee identifies shared definitions of dangerous AI behavior as a potentially achievable first step.

President Donald Trump will host Chinese leader Xi Jinping in Washington on Sept. 24 with AI governance on the agenda. The meeting follows an earlier agreement to create an AI-risk channel that, according to foreign-policy expert Lily McElwee, has seen little implementation—putting practical follow-through ahead of any sweeping safety pact.

The White House said in May that Trump would welcome Xi to Washington in the fall after the leaders agreed to build what it called a constructive relationship based on strategic stability, fairness and reciprocity. Trade and other issues are also expected at the summit, but AI has become a distinct test of whether the two governments can coordinate while competing for technological advantage.

A narrow agenda for AI safety

McElwee, president and CEO of the Phoenix Committee on Foreign Relations, expects “very limited progress” on AI safety. She sees a broad, Cold War-style arms-control agreement as unlikely because of deep mistrust and the difficulty of verifying compliance.

Two steps McElwee considers tentatively achievable

  • A government-to-government channel or hotline for discussing AI risks and early safeguards.
  • Shared definitions of dangerous AI behavior, including removing human control from weapons systems or integrating AI into the nuclear command chain.

The dispute the summit is unlikely to settle

China’s chief concern is U.S. export controls on advanced chips, McElwee said, and she expects those controls to remain unchanged at the summit. That leaves a hard boundary around the diplomacy: the countries may be able to discuss preventing particular AI uses, but not resolve the computing-access dispute that shapes their wider rivalry.

A functioning channel would therefore be a modest outcome, not a settlement of the competition. Its immediate value would be a place for both governments to raise AI risks and early safeguards; whether the earlier commitment becomes operational is the more concrete question heading into Washington.

Sources

  1. whitehouse.govFact Sheet: President Donald J. Trump Secures Historic Deals with China, Delivering for American Workers, Farmers, and Industry
  2. abcnews.comA global AI safety strategy depends on US-China cooperation. They each see the other as the problem
  3. azfamily.comTrump-Xi summit: Don’t expect a big AI safety deal, expert says

Loading discussion...