PeopleofSuperintelligence

Public voice

Thomas Larsen

Thomas Larsen is an AI policy researcher and scenario writer. He earned a computer science and mathematics degree at the University of Michigan in 2022, researched vision for self-driving cars, wrote an undergraduate thesis on robustness, joined MATS cohorts in 2022 and 2023, and wrote an overview of technical alignment. After AI safety research at MIRI, he founded the Center for AI Policy in 2023, submitted a public comment on AI risk to the National AI Advisory Committee, and coauthored Safety Cases in 2024. He coauthored AI 2027 in 2025 and a CSET report on AI research automation in January 2026. In July 2026, he coauthored AI 2040: Plan A; as of October 2026, he is Research Lead at the AI Futures Project and lead author of the scenario for US policymakers.

Works
3
Badges
Links
Google Scholar and LinkedIn

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2022–2023

    University of Michigan and MATS

    At the University of Michigan, he earns a degree in computer science and mathematics in 2022, conducts vision research for self-driving cars, and writes an undergraduate thesis on robustness. He joins MATS in its Summer 2022 cohort with John Wentworth and Winter 2023 cohort with Nate Soares. He also writes a detailed overview of technical AI alignment.

  2. August 2023

    Founding the Center for AI Policy

    After conducting AI safety research at MIRI, he founds the Center for AI Policy in August 2023, a DC-based group advocating policies to mitigate catastrophic AI risks. He serves as its Executive Director.

  3. October 2023

    A comment to NAIAC

    In October 2023, he submits CAIP's public comment to the National AI Advisory Committee on advanced AI's national-security risks and possible policy responses, including compute governance, frontier AI regulation, and international cooperation.

  4. March 2024

    Safety cases for advanced AI

    In March 2024, he coauthors Safety Cases: Justifying the Safety of Advanced AI Systems as CAIP's Director of Strategy. The paper proposes structured arguments that AI developers can use to show their systems are safe.

  5. April 2025

    AI 2027

    In April 2025, AI 2027 credits him alongside Daniel Kokotajlo, Scott Alexander, Eli Lifland, and Romeo Dean. Larsen, Kokotajlo, Lifland, and Dean write the scenario content; Alexander rewrites the text. The scenario explores possible AI progress and its consequences through 2027.

  6. January 2026

    When AI Builds AI

    In January 2026, he coauthors CSET's When AI Builds AI report, which distills consensus and disagreement from a workshop on automating AI research and development.

  7. July 2026

    AI 2040: Plan A

    In July 2026, he coauthors AI 2040: Plan A with Daniel Kokotajlo, Eli Lifland, and three other authors. The scenario recommends government action that delays superintelligence until 2040.

  8. As of October 2026

    Research Lead at AI Futures

    As of October 2026, he is Research Lead at the AI Futures Project and lead author of AI 2040: Plan A, a scenario illustrating the project’s recommendation to US policymakers.

Career

  1. NowAI Futures ProjectResearch Lead
  2. 2023Center for AI PolicyFounder, Executive Director
  3. Until 2022University of MichiganVision researcher
  4. Machine Intelligence Research InstituteAI safety researcher

Education

  1. 2022University of MichiganDegree in computer science and mathematics

From the biography.

Videos

YouTube videos in which Thomas Larsen speaks or appears, newest first. A model chose them from YouTube search results. Key moments are from summaries that Gemini wrote. List updated 20 October 2026.

  1. How Many Narrow AIs Could Behave Like One Superintelligence - Daniel Kokotajlo and Thomas LarsenMachine Learning Street Talk · 8 September 2026 · 1 h 30 min

    Key moments 2:14 This interview on Machine Learning Street Talk features… 35:30 "Humans in the Cloud" vs. General Tools 58:50 Buying Time & Pacing Frontier Capabilities 1:11:01 Accounting for roughly 99% of large-scale frontier compute 1:28:04 The "Vibe Shift" & Adaptivity

Video data from YouTube API Services. By watching you agree to the YouTube Terms of Service. How we use YouTube data.

  1. Direct Form 990 pay in the latest returns of 39 nonprofits with AI people on this site is highest at $2.1M at the Allen Institute for AI (Ai2) (2024)Includes Thomas Larsen. IRS Form 990 compensation across 39 nonprofits with AI people on this site from 2012 to 2025: in the latest returns, Mozilla reports $6.1M including related-company pay (2024) and the Allen Institute for AI (Ai2) leads direct nonprofit pay at $2.1M (2024).
  2. Of 216 AI researchers who entered public life, 83 held advisory seats and 58 took public rolesIncludes Thomas Larsen. Of 216 AI researchers who entered public life, 83 held advisory seats and 58 took public roles. Another 97 people moved from public careers into AI organizations or research, with 62% of dated records starting from 2020 on.
  3. Universities had a net outflow of 2,068 moves in AI careersIncludes University of Michigan. Our estimates of career flows and median h-index across universities and companies in AI, including the role of joint appointments and earlier academic jobs.
  4. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Machine Intelligence Research Institute. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Thomas Larsen yet. Send us a free-licensed photo.

  1. AI Risk Mitigation Fund airiskfund.com. Checked 22 October 2026.
  2. Ep 11 - Technical alignment overview w/ Thomas Larsen podbean.com. Checked 22 October 2026.
  3. Thomas Larsen - MATS Program Alumni matsprogram.org. Checked 22 October 2026.
  4. (My understanding of) What Everyone in Technical Alignment is Doing and Why — LessWrong lesswrong.com. Checked 22 October 2026.
  5. Introducing the Center for AI Policy (& we're hiring!) — LessWrong lesswrong.com. Checked 22 October 2026.
  6. NAIAC Public Comment | Center for AI Policy | CAIP centeraipolicy.org. Checked 22 October 2026.
  7. Safety Cases: Justifying the Safety of Advanced AI Systems | Center for AI Policy | CAIP centeraipolicy.org. Checked 22 October 2026.
  8. AI 2027: What Superintelligence Looks Like — LessWrong lesswrong.com. Checked 22 October 2026.
  9. About — AI 2027 ai-2027.com. Checked 22 October 2026.
  10. When AI Builds AI | Center for Security and Emerging Technology cset.georgetown.edu. Checked 22 October 2026.
  11. AI 2040: Plan A blog.aifutures.org. Checked 22 October 2026.
  12. About — AI Futures Project aifutures.org. Checked 22 October 2026.