From Foreign Policy · James Palmer · · 2m
Why Trump and Xi may split on AI safety
A Washington summit may put AI safety on the table, but the U.S. and China see the risks in very different ways.
In brief
A Washington summit may put AI safety on the table, but the U.S. and China see the risks in very different ways. AI safety talks will struggle unless Washington and Beijing first agree on which harms they are trying to prevent. Originally reported by Foreign Policy.
Summit hopes are thin
Trump and Xi are set to meet in Washington on Sept. 24, with AI safety likely on the agenda. Hopes for real alignment remain low.
“the chances of productive alignment on AI safety are low”
Cold War analogies tempt
Some analysts want AI rules to follow the pattern of U.S.-Soviet cooperation on arms control, disease, and the ozone layer, even amid rivalry.
Taiwan noise clouds talks
China has reportedly threatened to cancel if U.S. arms sales to Taiwan proceed first. Palmer treats that as routine pressure, not the main obstacle.
“routine diplomatic noise”
They fear different AI
U.S. tech leaders fixate on superintelligence and machine control. Chinese officials focus more on technical failures, jobs, and political stability.
Shared risks still exist
Security failures could give both sides common ground. The OpenAI-Hugging Face breach showed how dependent rivals still are on shared digital systems.
Regulation is mismatched
China can force its tech firms into line with fines, blocked IPOs, ideology checks, and pressure on executives. The U.S. system gives companies more room to fight.
Congress limits promises
U.S. AI rules would likely need new legislation, hard to pass in a divided Congress. Beijing may read the gap between promises and law as bad faith.
Trump rejects guardrails
Trump has made his hostility to AI regulation plain, saying strong presidential control is enough and blaming China for opposition to data centers.
“the only control or ‘guardrails’ that AI needs”
Rhetoric may bite back
Trump may think anti-China talk at home can coexist with cooperation. Chinese leaders may not separate the two so easily.
Beijing sees information war
China’s state security chief described U.S. AI models as tools in a propaganda and cognitive war against China, hardening the mood before talks.
“propaganda war and a cognitive war”
Xi courts BRICS
Xi made a short visit to India for the BRICS summit, signaling some repair work with New Delhi after years of border strain.
Other pressures build
China is refilling oil stocks, which may lift prices and pressure Trump. A deadly Qingdao shipyard fire has also drawn a high-level investigation.
The short version
AI safety talks will struggle unless Washington and Beijing first agree on which harms they are trying to prevent.
