Observed arrival · 2026-10-08
AI 2040’s Plan A: A Slower Road to Superintelligence
A detailed scenario proposes delaying superintelligence until 2040 through international agreement, research transparency, and coordinated safeguards.
- For
- AI policy, security, and governance readers
- Worth noticing
- The authors compare Plans B, C, D, and S, and explicitly call Plan A a recommendation rather than their best-guess forecast.
Field notes
Navigation breaks the proposal into sections on covert projects, verification, transparency, capability scaling, security, compute, economics, and alignment. The authors say the scenario draws on conversations with experts at frontier AI companies, tabletop exercises, and discussions with policy and national-security specialists; these are described as inputs, not independently documented here. The page also links to an economic-growth explorer and comparisons of alternative plans.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue