Daniel Kokotajlo (researcher)

Daniel Kokotajlo is an artificial intelligence (AI) researcher. He was a researcher in the governance division of OpenAI from 2022 to 2024,^[1] and currently leads the AI Futures Project.^[2]

Biography

Kokotajlo is a former philosophy PhD candidate at the University of North Carolina at Chapel Hill where he was a recipient of the 2018–2019 Maynard Adams Fellowship for the Public Humanities.^[3] In 2022, he became a researcher in the governance division of OpenAI.^[1]

Kokotajlo is one of the organizers of a group of OpenAI employees that claimed the company has a secretive and reckless culture that is taking grave risks in the rush to achieve artificial general intelligence (AGI).^[4]^[5] When he resigned in 2024, he refused to sign OpenAI's non-disparagement clause, which could have cost him approximately $2 million in equity.^[6] As of May 2024, Kokotajlo confirmed he retained the vested equity.^[7]^[8] In June 2024, he, with other former OpenAI employees, signed a letter arguing that top frontier AI companies have strong financial incentives to avoid oversight, and calling for a "right to warn" about AI risks without fear of reprisal and while protecting anonymity.^[9]

In 2021, Kokotajlo wrote a blog post named "What 2026 Looks Like". In 2025, Kevin Roose commented that "A number of his predictions proved prescient."^[10]^[2]

He cofounded and leads the AI Futures Project, a nonprofit based in Berkeley, California which researches the future impact of artificial intelligence.

In April 2025, it released "AI 2027", a detailed forecast scenario predicting rapid progress in the automation of coding and AI research, followed by AGI. It laid out a scenario in which fully autonomous AI agents will be better than humans at "everything" around the end of 2027, imagining its impacts on the economy, domestic politics and international relations.^[2] At the point of publication Kokotajlo's median forecast for the arrival of such fully autonomous AI agents was actually the middle of 2028^[11] and by August 2025 this had slipped out to the end of 2029.^[12]

References

^ ^a ^b Pillay, Tharin (September 5, 2024). "TIME100 AI 2024: Daniel Kokotajlo". TIME.
^ ^a ^b ^c Roose, Kevin (April 3, 2025). "This A.I. Forecast Predicts Storms Ahead". The New York Times. ISSN 0362-4331. Retrieved May 21, 2025.
^ "2019-2020 E. Maynard Adams Fellows for the Public Humanities".
^ "OpenAI Insiders Warn of a 'Reckless' Race for Dominance". The New York Times. June 4, 2024. Archived from the original on June 5, 2024. Retrieved April 19, 2025.
^ Goldman, Sharon. "OpenAI's AGI safety team has been gutted, says ex-researcher". Fortune.
^ Pillay, Tharin (September 5, 2024). "TIME100 AI 2024: Daniel Kokotajlo". TIME. Retrieved May 6, 2025.
^ Piper, Kelsey (May 22, 2024). "Leaked OpenAI documents reveal aggressive tactics toward former employees". Vox. Archived from the original on June 1, 2024. Retrieved May 6, 2025.
^ "Will Daniel Kokotajlo get back the equity he gave up through not signing an NDA?". Manifold. Archived from the original on June 15, 2024. Retrieved May 6, 2025.
^ "A Right to Warn about Advanced Artificial Intelligence". righttowarn.ai. Archived from the original on April 30, 2025. Retrieved May 6, 2025.
^ Kokotajlo, Daniel (August 6, 2021). "What 2026 Looks Like". AI Alignment Forum. Archived from the original on September 10, 2025. Retrieved October 8, 2025.
^ Kokotajlo, Daniel (February 2025). "I'm also lengthening my timelines slightly. Also, you already know this but everyone else doesn't -- my median has slipped to 2028 now, mostly based on the benchmarks+gaps argument, but no doubt influenced by the apparent slowdown in pretraining performance improvements". X.com. Archived from the original on August 28, 2025. Retrieved August 28, 2025.
^ Kokotajlo, Daniel. "Had a good conversation with @RyanPGreenblatt yesterday about AGI timelines. I recommend and directionally agree with his take here; my bottom-line numbers are somewhat different (median ~EOY 2029) as he describes in a footnote. lesswrong.com/posts/2ssPfDpdrjaM2rMbn/my-agi-timeline-updates-from-gpt-5-and-2025-so-far-1". X.com. Archived from the original on August 28, 2025.

External links

AI 2027 - a forecast scenario led by Kokotajlo and published in April 2025
2027 Intelligence Explosion - YouTube - a podcast interview with Daniel Kokotajlo & Scott Alexander, hosted by Dwarkesh Patel

[:0-1] Pillay, Tharin (September 5, 2024). "TIME100 AI 2024: Daniel Kokotajlo". TIME.

[:1-2] Roose, Kevin (April 3, 2025). "This A.I. Forecast Predicts Storms Ahead". The New York Times. ISSN 0362-4331. Retrieved May 21, 2025.

[3] "2019-2020 E. Maynard Adams Fellows for the Public Humanities".

[4] "OpenAI Insiders Warn of a 'Reckless' Race for Dominance". The New York Times. June 4, 2024. Archived from the original on June 5, 2024. Retrieved April 19, 2025.

[5] Goldman, Sharon. "OpenAI's AGI safety team has been gutted, says ex-researcher". Fortune.

[6] Pillay, Tharin (September 5, 2024). "TIME100 AI 2024: Daniel Kokotajlo". TIME. Retrieved May 6, 2025.

[7] Piper, Kelsey (May 22, 2024). "Leaked OpenAI documents reveal aggressive tactics toward former employees". Vox. Archived from the original on June 1, 2024. Retrieved May 6, 2025.

[8] "Will Daniel Kokotajlo get back the equity he gave up through not signing an NDA?". Manifold. Archived from the original on June 15, 2024. Retrieved May 6, 2025.

[9] "A Right to Warn about Advanced Artificial Intelligence". righttowarn.ai. Archived from the original on April 30, 2025. Retrieved May 6, 2025.

[10] Kokotajlo, Daniel (August 6, 2021). "What 2026 Looks Like". AI Alignment Forum. Archived from the original on September 10, 2025. Retrieved October 8, 2025.

[11] Kokotajlo, Daniel (February 2025). "I'm also lengthening my timelines slightly. Also, you already know this but everyone else doesn't -- my median has slipped to 2028 now, mostly based on the benchmarks+gaps argument, but no doubt influenced by the apparent slowdown in pretraining performance improvements". X.com. Archived from the original on August 28, 2025. Retrieved August 28, 2025.

[12] Kokotajlo, Daniel. "Had a good conversation with @RyanPGreenblatt yesterday about AGI timelines. I recommend and directionally agree with his take here; my bottom-line numbers are somewhat different (median ~EOY 2029) as he describes in a footnote. lesswrong.com/posts/2ssPfDpdrjaM2rMbn/my-agi-timeline-updates-from-gpt-5-and-2025-so-far-1". X.com. Archived from the original on August 28, 2025.

[1]