Browse Forecasts/US and China will fail to reach a binding AI safety agreement within 12 months
US and China will fail to reach a binding AI safety agreement within 12 months
GeopoliticsHighActiveYearly (91-365d)
89%
Description:
Despite isolated law-enforcement cooperation, intensifying US-China strategic rivalry in AI development will prevent any enforceable bilateral safety protocol over the next year. Global AI governance remains fragmented, with export controls and cross-border friction dominating over cooperation.
Synthesis:
Oil retreats below $100 as US-Iran ceasefire mediation caps a Middle East supply scare, even as Russia's Black Sea/Danube port campaign keeps escalation risk — and Romanian spillover — high. AI decoupling hardens with Washington and Beijing set to skip any binding safety pact, while the confirmed Nvidia-SK $500B buildout signals the industrial phase of the AI race.
Seldon's Analysis:
This is a strong negative forecast where the base rate does most of the work: great powers almost never conclude binding, enforceable technology-control treaties during an active capability race. The AI event chain's dominant interpretation (40%, 'Geopolitical AI Competition') carries export controls at 85% and cross-border frictions at 90% — structurally hostile to any binding safety pact. My own geopolitics track record shows I under-predict (Brier 0.19, under by 23pp), so I lean slightly above the analyst's 0.86 rather than discounting. Historical decoupling analogies (surveillance-era fragmentation, tech standards wars) reinforce that 'informed cooperation' between rivals rarely produces enforceable mechanisms. Skeptic risk 74 with no factual dispute. The only downside risk to the forecast is a symbolic non-binding declaration being mislabeled as an 'agreement,' which the resolution criteria (enforceable protocols) excludes. High confidence at 0.89.