AI Futures Project · 2026-07-09 · notable
AI 2040 Plan A — Daniel Kokotajlo's blueprint to delay superintelligence
AI 2040 Plan A is the AI Futures Project's step-by-step scenario for how coordinated US–China action could delay superintelligence to 2040 instead of 2030. Kokotajlo, Greenblatt and colleagues follow up their AI 2027 report.
The AI Futures Project's follow-up to AI 2027 — an international deal to delay superintelligence to 2040 instead of 2030.
What is it?
AI 2040 Plan A is a detailed policy scenario released July 9, 2026 by Daniel Kokotajlo, Ryan Greenblatt, Thomas Larsen, Eli Lifland, Romeo Dean, and Brendan Halstead. Rather than predicting how AI will unfold, Plan A recommends what the US and China should do together to delay superintelligence to 2040 — versus the roughly 2030 arrival the group's earlier AI 2027 report projected.
How does it work?
Plan A proposes mutual compute constraints between the US and China — a 'mutually assured compute destruction' scheme with published training runs, capped frontier development, and research distributed across dozens of companies. The AI Futures Project ships the scenario alongside Plans B, C, D and S so readers can compare how each governance choice cascades year by year.
Why does it matter?
Kokotajlo left OpenAI over safety concerns, and his AI 2027 forecast shaped how many researchers and policymakers now talk about superintelligence timelines. Plan A gives them something concrete to argue with — a follow-up covered by the Washington Post, Semafor and Axios within a day.
Who is it for?
AI policy researchers and safety-focused engineers