Daniel Kokotajlo
Executive director, AI Futures Project
Who is Daniel Kokotajlo
Daniel Kokotajlo is the executive director of the AI Futures Project, a small research group that forecasts the future of AI[1][5]. He worked as a governance researcher at OpenAI from 2022 to April 2024, where he focused on scenario planning[11][12][3]. He co-wrote AI 2027, a scenario of how AI could develop through 2027, published in April 2025[4][2].
TIME named him in its TIME100 AI lists for 2024 and 2025[11][17]. He is a different person from Daniel Kokotajlo the British film director, who made Apostasy and Starve Acre[21].
Philosophy and early AI forecasting
Kokotajlo earned a bachelor's degree in philosophy from the University of Notre Dame in 2014 and a master's in philosophy from the University of North Carolina at Chapel Hill in 2017[8]. He then started a philosophy PhD and left it to work at AI Impacts, and later at the Center on Long-Term Risk[1]. UNC's philosophy department named him a Maynard Adams Fellow for the Public Humanities for 2019–2020[9].
In August 2021 he posted "What 2026 Looks Like", a year-by-year scenario for 2022 to 2026, on the AI Alignment Forum[10]. TIME wrote in 2025 that it predicted the rise of chatbots and "inference-time scaling"[17].
Work at OpenAI and leaving
Kokotajlo joined OpenAI as a governance researcher in 2022 and left in April 2024[11][12]. In his own words, he quit "due to losing confidence that it would behave responsibly around the time of AGI"[1].
When he left, he refused to sign OpenAI's non-disparagement agreement[11][13]. On LessWrong on 10 May 2024 he wrote that the equity he gave up "probably would have been about 85% of my family's net worth at least"[13]. TIME put the sum at about $2 million and wrote that he put his freedom to speak up first[11].
In May 2024 OpenAI told former employees in a memo that it "has not canceled, and will not cancel, any vested units", whether or not they had signed the agreement[15][16].
On 4 June 2024 he and other current and former OpenAI and Google DeepMind staff published "A Right to Warn about Advanced Artificial Intelligence"[14]. The letter asks AI companies not to enforce agreements that bar criticism about risks, and not to withhold vested benefits in retaliation[14]. It also asks for an anonymous channel to boards, regulators and outside experts, a culture of open criticism, and no retaliation against staff who go public after other channels fail[14]. Yoshua Bengio, Geoffrey Hinton and Stuart Russell endorsed it[14].
The AI Whistleblower Initiative (AIWI) published a case study on his departure and the Right to Warn letter[12]. It says Harvard Law professor Lawrence Lessig approached Kokotajlo after Vox reported on OpenAI's exit agreements, and then represented "the whistleblowers" pro bono[12].
Profile of Leopold Aschenbrenner, who also left OpenAI in April 2024
AI Futures Project and AI 2027
The AI Futures Project describes itself as a small research group forecasting the future of AI, funded by charitable donations and grants[5]. Its core team includes Kokotajlo, Eli Lifland, Thomas Larsen and Romeo Dean[5]. As executive director, Kokotajlo oversees its research and policy recommendations[2].
AI 2027 was the group's first project, published on 3 April 2025 at ai-2027.com[4][22]. It is a detailed scenario forecast: AI agents arrive in 2026, coding is fully automated in early 2027, and an intelligence explosion follows in late 2027[4]. It has two endings, a "race" ending in which AI takes over and a "slowdown" ending[4][22]. Kokotajlo, Lifland, Larsen and Dean wrote it, and the writer Scott Alexander rewrote it in a more engaging style[2].
TIME reported that Kokotajlo co-wrote it after hosting a series of war games with participants from leading AI labs[17]. TIME wrote that he cautions the scenario is "a best guess at where the world might be headed; one possible path among many"[17]. TIME reported that US Vice President J.D. Vance said he had read it[17].
How his forecasts changed, AI 2040 and the 2026 update
Kokotajlo has changed his AGI forecast in public several times[18]. His median year for AGI was 2027 from 2022, and in February 2025, before AI 2027 came out, he moved it to 2028[18]. By August 2025 it was about the end of 2029, and by November 2025 "around 2030, lots of uncertainty though"[18]. The group's December 2025 model update pushed full automation of coding into the early 2030s[19][8].
On 9 July 2026 he and his co-authors published "AI 2040: Plan A"[20]. They call it "a recommendation, not a prediction"[20]. In it, the US and Chinese governments take decisive action and delay the creation of superintelligence to 2040[20].
On 30 September 2026 he testified before the US Senate Subcommittee on Disaster Management, District of Columbia, and Census[7]. He asked for more transparency into the AI industry and for less computing power to go to automating AI research[7]. He said AI company leaders had agreed on the need to "pace the frontier" but had not done so, and that the US government should step in[7].
On 10 November 2026 the group published a forecast update titled "Looks Like the Intelligence Explosion Will Happen in 2027 After All"[6]. The group estimated that a METR report implies AI is speeding up AI progress by about 1.7 times[6]. All the authors now forecast that a superhuman AI researcher will likely arrive in 2027, unless AI development is deliberately slowed[6]. Kokotajlo moved his median for top-expert-dominating AI to September 2027, from late 2028[6]. The update says that if this is right, "it is incredibly urgent to pause or substantially 'pace' frontier AI development"[6].
Summary of his Joe Rogan Experience interview (#2551)
Known for
- Leading the AI Futures Project and co-writing AI 2027
- Refusing OpenAI's non-disparagement agreement in 2024
- The Right to Warn letter on AI company employees' freedom to speak
Frequently asked questions
Who is Daniel Kokotajlo?
He is the executive director of the AI Futures Project and a co-author of AI 2027. He was a governance researcher at OpenAI from 2022 to April 2024. He is not the British film director of the same name.
What is AI 2027?
AI 2027 is a scenario forecast published on 3 April 2025 by the AI Futures Project. It describes AI agents in 2026, fully automated coding in early 2027 and an intelligence explosion in late 2027, with a "race" ending and a "slowdown" ending.
What is the AI Futures Project?
It is a small research group that forecasts the future of AI, funded by charitable donations and grants. Kokotajlo is its executive director, and its team includes Eli Lifland, Thomas Larsen and Romeo Dean.
Did Daniel Kokotajlo give up his OpenAI equity?
He refused to sign OpenAI's non-disparagement agreement when he left in 2024, putting at risk equity that TIME put at about $2 million. He wrote that it was probably about 85% of his family's net worth. In May 2024 OpenAI told former employees it would not cancel any vested units, whether or not they had signed.
Why did Daniel Kokotajlo leave OpenAI?
He says he quit because he lost confidence that OpenAI would behave responsibly around the time of AGI. In June 2024 he and other former and current staff published the Right to Warn letter.
When does Daniel Kokotajlo now expect superhuman AI?
In the AI Futures Project's update of 10 November 2026, he and his co-authors forecast that a superhuman AI researcher will likely arrive in 2027, unless AI development is deliberately slowed. His median for top-expert-dominating AI moved to September 2027.
Sources
- LessWrong: Daniel Kokotajlo profile — checked 12 Nov 2026
- AI 2027: About — checked 12 Nov 2026
- Apple Podcasts: The Joe Rogan Experience #2551, Daniel Kokotajlo, 9 September 2026 — checked 12 Nov 2026
- AI Futures Project: Our first project, AI 2027, 3 April 2025 — checked 12 Nov 2026
- AI Futures Project — checked 12 Nov 2026
- AI Futures Project: Q3.5 timelines update, 10 November 2026 — checked 12 Nov 2026
- AI Futures Project: Senate testimony, 2 October 2026 — checked 12 Nov 2026
- Wikipedia: Daniel Kokotajlo — checked 12 Nov 2026
- UNC Philosophy: Graduate students to serve as Adams Fellows, October 2019 — checked 12 Nov 2026
- AI Alignment Forum: What 2026 Looks Like, August 2021 — checked 12 Nov 2026
- TIME100 AI 2024: Daniel Kokotajlo — checked 12 Nov 2026
- AIWI: Right to Warn and Daniel Kokotajlo — checked 12 Nov 2026
- LessWrong: Daniel Kokotajlo's comment on the equity, 10 May 2024 — checked 12 Nov 2026
- A Right to Warn about Advanced Artificial Intelligence, 4 June 2024 — checked 12 Nov 2026
- CNBC via NBC Los Angeles: OpenAI sends internal memo releasing former employees from exit agreements, 23 May 2024 — checked 12 Nov 2026
- Engadget: OpenAI scraps its non-disparagement agreement, 24 May 2024 — checked 12 Nov 2026
- TIME100 AI 2025: Daniel Kokotajlo — checked 12 Nov 2026
- AI Futures Project: How our AI timelines forecasts have changed since AI 2027, 27 January 2026 — checked 12 Nov 2026
- AI Futures Project: AI Futures Model, December 2025 update — checked 12 Nov 2026
- AI Futures Project: AI 2040: Plan A, 9 July 2026 — checked 12 Nov 2026
- IMDb: Daniel Kokotajlo, film director — checked 12 Nov 2026
- AI 2027 — checked 12 Nov 2026