A Substack newsletter by Ajeya Cotra exploring AI capabilities, timelines, and the societal implications of increasingly autonomous AI systems.
A Substack newsletter by Ajeya Cotra exploring AI capabilities, timelines, and the societal implications of increasingly autonomous AI systems.
People
Updated 05/18/26Editor and primary author
Co-author of early blog version
Funding Details
Updated 05/18/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26Planned Obsolescence is a Substack newsletter written by Ajeya Cotra, a researcher whose work centers on understanding and preparing for the risks posed by extremely powerful AI systems. Cotra previously worked at Open Philanthropy, where she led grantmaking on technical AI alignment research and produced influential work on AI timelines and the risks of training AI systems to be deceptive. She now works at METR, an organization focused on evaluating frontier AI capabilities and risks. The newsletter was originally launched in March 2023 as a collaborative blog with Kelsey Piper, a senior writer at Vox's Future Perfect. That version, hosted at planned-obsolescence.ghost.io, included seven initial posts on AI alignment, situational awareness, and researcher disagreements, along with AI-generated audio narration. The current version, relaunched on Substack in September 2024 at www.planned-obsolescence.org, is edited and primarily written by Cotra alone. The newsletter's tagline is "Thinking ahead to a future where AI decides everything." It publishes occasionally and covers topics including AI capabilities benchmarks, proposed milestones for AI progress (such as "self-sufficient AI"), takeoff speed scenarios, forecasting AI's role in R&D and economic competition, and implications for AI safety research priorities. Posts are frequently crossposted to the EA Forum and LessWrong and have a wide readership within the AI safety and effective altruism communities. As of early 2026, the newsletter has over 2,000 free subscribers. Paid subscriptions are disabled; the newsletter is free to read. Recent posts include analyses of AI capabilities forecasting errors, 2026 AI predictions, and critiques of AGI definitions.
Theory of Change
Updated 05/18/26By publishing accessible, carefully reasoned analyses of AI capabilities and timelines, Cotra aims to help researchers, policymakers, funders, and the broader EA community develop accurate models of how and when AI systems might become transformatively powerful or dangerous. Better-calibrated beliefs about AI progress are expected to improve research prioritization, policy decisions, and grantmaking in AI safety — ultimately increasing the probability that the transition to powerful AI goes well for humanity.
Grants Received– no grants recorded
Updated 05/18/26Projects
Updated 05/18/26Podcast feed providing AI-generated audio versions of Planned Obsolescence posts from planned-obsolescence.org, read by models trained on the authors’ voices.
Discussion
No comments yet. Be the first to share your thoughts.