Daniel Kokotajlo
Daniel Kokotajlo directs research at the nonprofit AI Futures Project. His public work uses detailed scenarios and models to examine possible AI development. He wrote What 2026 Looks Like in 2021 and later worked at OpenAI on forecasts and evaluations, according to his testimony. He resigned in 2024 and described concerns about the company's conduct and restrictions on criticism. Those departure reasons are his account. He signed A Right to Warn and coauthored AI 2027 with four colleagues. His later work includes the AI Futures Model, revised timeline methods, and AI 2040: Plan A. The forecasts depend on assumptions and their accuracy remains unknown. AI 2040 principally presents policy recommendations, while his Senate testimony states his proposed responses.
- Now at
- AI Futures Project
- Works
- 6
- Citations
- 69
- h-index
- 3
Publication numbers from OpenAlex, as of 14 October 2026.
Timeline
-
August 2021
Writing a possible future
He publishes a possible future history to make his assumptions about AI explicit.
-
2024
Leaving and speaking publicly
He resigns from OpenAI and publicly describes his concerns.
-
April 2025
A team writes AI 2027
He and four colleagues publish AI 2027, a scenario with predictive intent.
-
2025–2026
Making the assumptions adjustable
At the AI Futures Project, he contributes ideas and feedback to a shared forecasting model.
-
September–October 2026
Turning scenarios into policy discussion
He coauthors a policy scenario and calls for transparency in Senate testimony.
Sources and credits
Photo: Daniel Kokotajlo; image supplied by www.aifutures.org. License: Not stated by source. Source: aifutures.org. Cropped to a square and resized. Ask us to remove or replace this photo.
- AI Futures Project: current team
- What 2026 Looks Like, 6 August 2021
- Kokotajlo: Senate testimony post, 2 October 2026
- Kokotajlo: written Senate testimony
- A Right to Warn, 4 June 2024
- AI 2027: original scenario
- AI Futures Model: contributors and changelog
- Q2.5 2026 Timelines Update, 16 August
- AI 2040: Plan A