AI Futures Project releases Plan A, a U.S.-China coordination framework for advanced AI development through mutual compute monitoring and regulatory oversight
Policy change ● Confirmed 78% confidence first seen
The AI Futures Project released Plan A, a detailed roadmap proposing how the United States and China could jointly manage advanced AI development through the 2040s. The plan centers on establishing a shared regulatory regime with mutual control over chip supply chains and transparent data centers, verified through mutual auditors monitoring 98.5% of existing AI computing hardware. The proposal, authored by prominent AI forecasters including Daniel Kokotajlo and Ryan Greenblatt, has drawn mixed reactions including criticism from those skeptical of superintelligence timelines and compute restriction advocates.
Decision brief
- What changed
- The AI Futures Project (authors of the AI 2027 forecasts, including Daniel Kokotajlo and Ryan Greenblatt) published 'Plan A,' a proposed U.S.-China coordination framework for managing advanced AI development through the 2040s, centered on mutual chip-supply control, transparent data centers, and mutual auditors monitoring 98.5% of existing AI compute hardware.
- Why it matters
- This is a think-tank proposal, not enacted policy, but it signals how influential AI forecasters envision future compute governance and export-control regimes that could eventually shape chip supply chains, data center transparency requirements, and cross-border auditing obligations. Companies with significant compute infrastructure or China-exposed supply chains should track whether this framework gains traction with policymakers, as early positioning could matter if similar verification regimes become policy.
- Evidence
- The plan is covered consistently across three sources (TLDR summary and two Zvi/Don't Worry About the Vase posts), all describing the same core mechanism and authorship, but all coverage traces back to the AI Futures Project's own release rather than independent government or industry confirmation of adoption.
- What remains uncertain
- It is unclear whether any government—U.S., China, or otherwise—has expressed interest in adopting Plan A, or whether the 98.5% compute monitoring target is technically feasible; the plan has already drawn criticism from those skeptical of superintelligence timelines and from compute-restriction advocates, suggesting contested assumptions underlying the proposal. There is no evidence in the coverage of official government response.
- Monitor next
- Watch for any formal response or engagement from U.S. or Chinese government officials, or from major AI labs/chipmakers, regarding Plan A's compute-monitoring and auditing mechanisms.
Analytical support, not advice — assumptions and open questions stated above.