Why Oxford's Future of Humanity Institute Closed in 2024

A freeze that never lifted

The University of Oxford's Faculty of Philosophy froze FHI's fundraising and hiring starting in 2020. The freeze cost the institute lead researchers and, in the institute's own account, an especially promising cohort of junior researchers who left for other organizations. In late 2023 the Faculty decided the contracts of the remaining staff would not be renewed, and on April 16, 2024 the institute was closed. [3][1]

FHI's own closing statement, still on its website, attributes the end to "increasing administrative headwinds within the Faculty of Philosophy" rather than to any single decision. A 404 Media report published the day after closure described staff learning the institute was done when university IT arrived to collect their computers. [1][4]

Read against the final report, the 2020 freeze looks like the effective end of FHI, with the 2023 non-renewal and the April 2024 closure only making that outcome formal four years later. Nothing in the sources points to a single triggering event; it reads as an institution slowly starved of the ability to hire or raise money inside a faculty structure it had outgrown. [3]

What it left behind

Nick Bostrom founded FHI in 2005 inside Oxford's Faculty of Philosophy, funded initially through the Oxford Martin School (then the James Martin 21st Century School) and incubated alongside the Oxford Uehiro Centre for Practical Ethics. FHI's account credits the institute with helping develop the vocabulary and early research programs behind existential risk, longtermism, effective altruism, AI alignment and AI governance. [3][1]

Bostrom wrote Superintelligence: Paths, Dangers, Strategies while directing FHI, and the final report describes the institute's later years spent on AI risk and governance work, a digital minds program, and a Grand Futures research line on the limits of what future intelligence could achieve. [3]

FHI never shipped a product in the sense most entries on this timeline do. Its clearest legacy is conceptual: terms and framings that now read as background assumptions in AI safety and governance discourse were coined or systematized inside this one small, informally run institute before larger, better funded labs and think tanks took the same questions on. [3][1]

An oral history of itself

Anders Sandberg, one of FHI's longest serving researchers, wrote the institute's closing document, deposited with Oxford's research archive as "Future of Humanity Institute 2005-2024: Final Report." Its preamble describes the reversal directly: normally a manifesto comes first and stimulates the work that follows, while this document is "an epitaph summarizing what the Future of Humanity Institute was, what we did and why, what we learned, and what we think comes next," which Sandberg calls "an oral history of FHI from some of its members." [3]

The 92 page report includes a quote from Bostrom praising the administrative staff who kept the institute running inside Oxford's bureaucracy for almost two decades, singling out individuals by name for work he says even FHI's own researchers did not fully appreciate at the time. [3]

Sandberg flags the document's own limits in the preamble: it will not be unbiased or complete, and other former members would place the emphasis differently. It is a former insider's account written after the fact rather than an independent history, which makes it a valuable primary source on how FHI understood itself and a poor substitute for outside verification of any specific claim inside it. [3]

The Oxford Martin School’s historical program page dates FHI’s founding to 2005 and says the School’s direct funding ended in 2012. Its present-tense wording is an institutional archive, rather than evidence that FHI still operates. [2]