Seguire
- X@DKokotajlo
- Substackblog.aifutures.org
- Sito webaifutures.org
- Wikipediaen.wikipedia.org
Daniel Kokotajlo
Ex ricercatore di governance di OpenAI, dimessosi nel 2024 per come l’azienda gestiva il rischio dell’AGI, oggi guida l’AI Futures Project, i cui scenari AI 2027 e AI 2040 avvertono che la superintelligenza è vicina e chiedono un accordo internazionale per ritardarla.
Ruoli
- 2022 – apr 2024Researcher, governance team, OpenAIResigned in April 2024 and declined to sign a non-disparagement clause, forfeiting vested equity.Fonte →
- ott 2024 – oggiExecutive Director, AI Futures ProjectNonprofit forecasting organization in Berkeley, California; published AI 2027 (April 2025) and AI 2040: Plan A (July 2026).Fonte →
A registro
29 set 2026 · Intervista · frominside.ai
This is exactly as dangerous as it sounds and must not be allowed to happen.
Executive Director, AI Futures Project all’epoca
From his filmed interview for frominside.ai, a Palisade Research project of interviews with current and former frontier-lab employees, at 0:53, under the project's question “Do they really believe AI could kill everyone?”. Date is the day the project was published and Reuters reported it; the day of filming is not stated.
9 set 2026 · Podcast · The Joe Rogan Experience
And then, you know, maybe they kill everyone and maybe they don't deliberately kill anyone, but they just use our habitat for some other type of infrastructure like more data centers or whatever.
Executive Director, AI Futures Project all’epoca
Episode #2551, released 9 September 2026 (https://podcasts.apple.com/us/podcast/2551-daniel-kokotajlo/id360084272?i=1000788718147); words as printed by HuffPost, whose caption also dates the episode 9 September. The audio was not checked; an automated transcript (https://podcasts.happyscribe.com/the-joe-rogan-experience/2551-daniel-kokotajlo) renders the same sentence without "you know,". On the same HuffPost page: "We're basically on a track to put them in charge of basically everything."
Ex AI Researcher Warns Joe Rogan Artificial Intelligence Could Turn On Humans And 'Kill Everyone' →
8 set 2026 · Post · Social media
They can't even align or control their AIs today, and the AIs of the future will be much smarter and harder to understand. Even if they succeed, they will then have created the most intense concentration of power in human history — a new artificial species of superintelligences obedient to the commands and values of a tiny handful of people. … This situation is insane and must not be allowed to continue.
Executive Director, AI Futures Project all’epoca
Written the day of Jacob Coxon's resignation, as quoted by CBS News, which dates the post 8 September but does not name the platform; the first two sentences are printed as one quotation, the third as an addition, hence the ellipsis. The original post is not fetchable; the wording is reproduced identically by AOL and Symplexia.
Humans are "close to being outsmarted" by superintelligence, AI expert says →
9 lug 2026 · Articolo scientifico · AI Futures Project
the least bad plan we currently know of
Executive Director, AI Futures Project all’epoca
A recommendation, not a prediction, that the US and China act to delay superintelligence to 2040 instead of the roughly 2030 arrival the group expects by default.
lug 2026 · Podcast · 80,000 Hours Podcast
The trends seem to indicate that we're just a couple years away from fully automating AI research — and that after that, superintelligence is probably not that far away.
Executive Director, AI Futures Project all’epoca
Recorded 27–28 July 2026.
27 gen 2026 · Post · AI Futures Project blog
Things seem to be going somewhat slower than the AI 2027 scenario. Our timelines were longer than 2027 when we published and now they are a bit longer still; 'around 2030, lots of uncertainty though' is what I say these days.
Executive Director, AI Futures Project all’epoca
Co-written with Eli Lifland and Brendan Halstead; gives his AGI median as December 2030.
Clarifying how our AI timelines forecasts have changed since AI 2027 →
3 apr 2025 · Articolo scientifico · AI Futures Project
We predict that the impact of superhuman AI over the next decade will be enormous, exceeding that of the Industrial Revolution.
Executive Director, AI Futures Project all’epoca
Scenario forecast co-authored with Scott Alexander, Thomas Larsen, Eli Lifland and Romeo Dean.
apr 2024 · Post · LessWrong
Quit OpenAI due to losing confidence that it would behave responsibly around the time of AGI.
Researcher, governance team, OpenAI all’epoca
His profile bio, written on leaving OpenAI; in a May 2024 comment he added that he refused the exit paperwork because he "wanted to retain my ability to criticize the company in the future" (https://www.lesswrong.com/posts/kovCotfpTFWFXaxwi/simeon_c-s-shortform).