Effective Altruism
Last reviewed
Apr 28, 2026
Sources
40 citations
Review status
Source-backed
Revision
v1 · 3,836 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Last reviewed
Apr 28, 2026
Sources
40 citations
Review status
Source-backed
Revision
v1 · 3,836 words
Add missing citations, update stale details, or suggest a clearer explanation.
Effective altruism (often abbreviated EA) is a philosophical and social movement that uses evidence and careful reasoning to identify the most effective ways to benefit others, then takes action on what those investigations recommend. The phrase was coined in 2011 by members of two Oxford-based projects, Giving What We Can and 80,000 Hours, and the underlying approach treats charitable giving, career choice, and policy work as problems amenable to rigorous comparative analysis rather than as expressions of personal preference.[1][2]
In practice, EA combines a moral premise (that all suffering deserves equal consideration regardless of where or when it occurs) with a methodological commitment (that some interventions do many times more good per dollar or per hour than others, and the difference is worth measuring). Adherents have channelled billions of dollars toward causes ranging from anti-malaria bednets in sub-Saharan Africa to research on risks from advanced artificial intelligence. By the mid-2020s the movement had become one of the dominant funding sources for AI safety and existential risk research, a development that has drawn both fresh interest and sustained criticism.[3][4]
The philosophical roots of effective altruism go back at least to the utilitarian tradition of Jeremy Bentham and John Stuart Mill, but the proximate intellectual ancestor is Peter Singer's 1972 essay "Famine, Affluence, and Morality," published in the journal Philosophy & Public Affairs. Singer argued that affluent people in wealthy countries are morally required to give substantially more to relieve global poverty than convention treats as virtuous, illustrating the point with the analogy of a child drowning in a shallow pond whom a passerby could save at small cost.[5] Singer's 2009 book The Life You Can Save extended the argument and helped seed the donor base that would later support EA organizations, including inspiring Dustin Moskovitz and Cari Tuna to direct their fortune toward high-impact philanthropy.[6]
The organisational core of the movement coalesced in Oxford between 2009 and 2012. In November 2009, ethicist Toby Ord launched Giving What We Can together with his wife Bernadette Young and the philosopher William MacAskill, then a graduate student. The society asked members to take a public pledge to donate at least 10 percent of their income to charities they believed to be among the most effective at improving lives. Ord himself went further, capping his own annual spending at roughly £20,000 and pledging the remainder of his earnings.[7][8]
In 2011, MacAskill and another Oxford graduate, Benjamin Todd, founded 80,000 Hours, named for the rough number of hours a person spends working over a career. The organisation's premise was that for ethically motivated people, choice of profession was probably a larger lever than choice of charity. Late that year, members of the two projects met to consider a name for the umbrella entity that would house both. They chose "Centre for Effective Altruism," and this was the first use of the phrase "effective altruism" in its current sense; before then, terms like "optimal philanthropy" and "rational altruism" had circulated. The Centre for Effective Altruism (CEA) was formally incorporated in 2012.[1][9]
A parallel strand grew up in the United States. In 2007, Holden Karnofsky and Elie Hassenfeld, two former analysts at the hedge fund Bridgewater Associates, raised about $300,000 from former colleagues and founded GiveWell, a charity evaluator that aimed to rank organisations by lives saved or improved per dollar. GiveWell's early reports, focused on global health programmes such as bednet distribution and deworming, set the template for the sort of cost-effectiveness analysis that became EA orthodoxy.[10][11] Around the same time, the Future of Humanity Institute at Oxford, founded by Nick Bostrom in November 2005, was developing the conceptual vocabulary, including "existential risk" and "longtermism," that would later orient much of EA's attention toward the long-run future.[12]
Effective altruism is not a single doctrine, but a working consensus has formed around a small set of methodological commitments.
Cause neutrality is the idea that resources should go to whichever problem will yield the most good per unit of effort, regardless of how the problem maps onto existing identity, geography, or sentimental attachment. A cause-neutral donor is in principle willing to switch from, say, funding scholarships in their hometown to funding a deworming programme in Tanzania if the second produces more welfare per dollar.[13]
Cost-effectiveness treats interventions as comparable on a common scale, often something like quality-adjusted life years (QALYs) saved per dollar or comparable welfare metrics for non-human animals. Empirical work in this area, much of it pioneered by GiveWell, has found order-of-magnitude differences between charities working on superficially similar problems.
Counterfactual impact asks not how much good resulted overall but how much would not have happened without the donor's contribution. A donation that merely substitutes for funding another donor would have provided counts for less than one that closes a real gap.
Scope sensitivity is the principle that helping ten thousand people is roughly ten times as valuable as helping one thousand, all else being equal. Psychological research suggests humans intuitively flatten this scale (a phenomenon called scope insensitivity), which EA practice tries to correct through quantitative reasoning.
Expected-value reasoning asks decision makers to weight outcomes by their probabilities. This commits EA-aligned funders to taking small probabilities of large outcomes seriously, which is one root of the movement's interest in low-probability catastrophes such as engineered pandemics or misaligned AGI.[14]
These principles are usually paired with a triage framework that scores potential causes on three axes: scale (how big is the problem?), tractability (how much progress can additional resources buy?), and neglectedness (how many other actors are already working on it?).[3]
EA-aligned funders and researchers tend to cluster around four broad cause areas, with significant overlap and disagreement about how to weight them.
The earliest and best-funded EA cause is global health, where evidence is strong and unit costs comparatively cheap. GiveWell's recommendations have steered hundreds of millions of dollars toward a small list of charities. In 2024 alone, GiveWell directed about $397 million to its top recommendations, with $150 million going to the Against Malaria Foundation, $87 million to the Malaria Consortium for seasonal malaria chemoprevention, $54 million to Helen Keller International for vitamin A supplementation, and $12 million to New Incentives for conditional cash transfers tied to childhood vaccinations.[15] Other EA-favoured organisations include GiveDirectly, which sends unconditional cash transfers to extremely poor households in East Africa, and Evidence Action's Deworm the World programme, which costs roughly $1 per child treated.[15][16]
A second focus is the welfare of farmed animals. Following Peter Singer's earlier work on animal liberation, EA-aligned analysts argue that the sheer numerical scale of factory farming (tens of billions of land animals slaughtered annually, plus orders of magnitude more aquatic animals) and the comparative neglect of the issue make it one of the highest-leverage causes for charitable spending. Funded organisations include corporate-campaign groups pushing for cage-free egg production, the Good Food Institute (alternative proteins), and Animal Charity Evaluators, which itself was incubated under 80,000 Hours before spinning off.[17]
The most contested cause area is longtermism, the view that influencing the long-run future is a leading moral priority. Practical work in this area concentrates on a short list of catastrophic risks: misaligned advanced AI, engineered pandemics, nuclear war, and runaway climate change. EA-aligned organisations working in the space include MIRI, the Center for AI Safety, Redwood Research, the Nuclear Threat Initiative, and biosecurity programmes at Johns Hopkins and elsewhere.[12][18]
A fourth category, sometimes called "meta-EA," funds the movement itself: outreach, conferences, fellowship pipelines, the EA Forum, and infrastructure for vetting other grants. Critics inside and outside the movement have argued that meta-funding can become circular, with EAs hiring other EAs to recruit yet more EAs, and the appropriate share of meta-spending remains an internal debate.
EA today is a network of dozens of nonprofits and funds. The following table summarises the most important.
| Organisation | Primary focus | Founded | Founders / lead figures |
|---|---|---|---|
| Giving What We Can | 10 percent giving pledge | 2009 | Toby Ord, William MacAskill, Bernadette Young |
| 80,000 Hours | High-impact careers research | 2011 | William MacAskill, Benjamin Todd |
| Centre for Effective Altruism | Movement infrastructure | 2012 | William MacAskill, Toby Ord |
| GiveWell | Charity evaluation, global health | 2007 | Holden Karnofsky, Elie Hassenfeld |
| Open Philanthropy | Large-scale grantmaking | Spun out 2017 | Cari Tuna, Dustin Moskovitz, Holden Karnofsky |
| Good Ventures | Family foundation funding Open Phil | 2011 | Cari Tuna, Dustin Moskovitz |
| Future of Humanity Institute | Existential risk research (closed 2024) | 2005 | Nick Bostrom |
| MIRI | Technical AI alignment | 2000 (as SIAI) | Eliezer Yudkowsky |
| Center for AI Safety | AI risk research and advocacy | 2022 | Dan Hendrycks, Oliver Zhang |
| Redwood Research | Empirical AI alignment | 2021 | Nate Thomas, Buck Shlegeris, Bill Zito |
| Longview Philanthropy | Major-donor advisory on emerging tech | 2018 | Simran Dhaliwal, Natalie Cargill |
| Forethought | Global priorities research, AI transition | 2018 | Founded by EA researchers in Oxford |
The biggest single funder of the EA ecosystem is Open Philanthropy, which began as a partnership between GiveWell and Cari Tuna and Dustin Moskovitz's family foundation Good Ventures, and became an independent limited liability company in June 2017. Cari Tuna is president, and Moskovitz, a co-founder of Facebook and CEO of Asana, is the principal funder.[19][20]
The link between EA and AI safety runs through both money and people, and tracking it requires distinguishing between three roles: funder, employer, and idea producer.
On the funding side, Open Philanthropy made one of the earliest large institutional grants to AI safety in March 2017, awarding $30 million to OpenAI as part of a board seat arrangement when OpenAI was still a nonprofit. The grant was controversial inside EA and is now sometimes cited as a cautionary example by the movement's critics. Open Phil has since become the largest single funder of technical AI alignment research, supporting work at Redwood, MIRI, the Center for AI Safety, the Alignment Research Center, and academic groups at Berkeley, Cambridge, and elsewhere.[21][22]
Anthropic, founded in May 2021 by siblings Dario and Daniela Amodei together with seven other former OpenAI staff, drew its initial backing from a roster heavy with EA-aligned investors. Its $124 million Series A round in 2021 was led by Jaan Tallinn, the Estonian co-founder of Skype and a long-standing donor to existential-risk causes; participants included Dustin Moskovitz, James McClave, the Center for Emerging Risk Research, and Eric Schmidt. A subsequent Series B round of about $580 million in 2022 was led by Sam Bankman-Fried, then a celebrated EA donor through his crypto exchange FTX. When FTX collapsed later that year, the Anthropic stake became part of the bankruptcy estate.[23][24]
On the talent side, 80,000 Hours since the late 2010s has explicitly named technical AI safety research as one of its top recommended career paths. The organisation has run workshops, an active job board, and a podcast that has placed many EAs into roles at AI labs, alignment-focused nonprofits, and government bodies. Critics inside and outside EA have argued that this pipeline has created a feedback loop between EA framing and AI lab strategy.[18]
Idea-wise, much of the philosophical scaffolding for taking AI risk seriously was built by people working in or adjacent to EA institutions: Eliezer Yudkowsky at MIRI, Nick Bostrom and Toby Ord at the Future of Humanity Institute, Stuart Russell at the Center for Human-Compatible AI at Berkeley, and Paul Christiano at the Alignment Research Center. Bostrom's 2014 book Superintelligence: Paths, Dangers, Strategies was a landmark in this regard, popularising the idea that misaligned advanced AI could pose a catastrophic risk.[12]
Longtermism is the moral view that positively influencing the long-run future is among the most important priorities of our time. Its philosophical roots go back to Bostrom's 2003 paper "Astronomical Waste: The Opportunity Cost of Delayed Technological Development," published in Utilitas, which argued that, given the enormous number of potentially happy lives that could exist in the future if humanity survives, even small reductions in existential risk have very high expected value.[25]
The view received its most rigorous philosophical statement in a 2021 working paper by Hilary Greaves and William MacAskill, "The Case for Strong Longtermism," published through the Global Priorities Institute at Oxford. Greaves and MacAskill distinguished axiological strong longtermism (far-future effects are the most important determinant of the value of our actions) from deontic strong longtermism (those effects are the most important determinant of what we ought to do). The paper argued that, on plausible empirical and ethical assumptions, optimising for the very long run dominates other considerations across a wide class of decisions.[26]
A looser "weak longtermism" merely claims that future people deserve substantial moral weight and that this weight is sometimes decisive, without claiming that long-run effects always dominate. MacAskill's 2022 book What We Owe the Future presented the case to a general audience, arguing that future people count, that there could be enormous numbers of them, and that humans alive today can take actions that improve their lives, especially by avoiding permanent catastrophes and by shaping the trajectory of long-lasting institutions.[27]
A companion text, Toby Ord's 2020 book The Precipice: Existential Risk and the Future of Humanity, focused specifically on the catastrophic-risk wing of the argument. Ord estimated a roughly one-in-six probability of an existential catastrophe within the next century, and assigned about one-in-ten of that risk to unaligned artificial general intelligence, more than all other sources combined in his estimate.[28]
In November 2022, the cryptocurrency exchange FTX, which had been founded in 2019 by Sam Bankman-Fried and was at one point valued at $32 billion, suffered a run on deposits and collapsed within days. Bankman-Fried, who had been one of the highest-profile new donors to EA and had committed substantial resources to its longtermist wing, resigned as chief executive on November 11, 2022. FTX, the trading firm Alameda Research, and roughly 130 affiliated entities filed for Chapter 11 bankruptcy the same day.[29][30]
The FTX Future Fund, which Bankman-Fried had established in February 2022 with a public commitment of around $160 million for 2022 alone (and a stated ambition of $1 billion per year in future grants), had concentrated much of its giving in AI safety, biosecurity, and EA movement infrastructure. Roughly $32 million had been disbursed for AI safety projects between February and August 2022 before the fund effectively shut down. Its entire board, which included William MacAskill, resigned on November 10, 2022, citing fundamental questions about FTX's legitimacy.[31]
Bankman-Fried was extradited from the Bahamas in December 2022, tried in federal court in Manhattan in October and November 2023, and on November 2, 2023 found guilty on all seven counts (two counts of wire fraud, two of conspiracy to commit wire fraud, conspiracy to commit securities fraud, conspiracy to commit commodities fraud, and conspiracy to commit money laundering). On March 28, 2024, Judge Lewis Kaplan sentenced him to 25 years in federal prison and ordered forfeiture of about $11 billion. The bankruptcy estate later sought to claw back charitable gifts from FTX-affiliated entities, including grants made to EA-aligned organisations, on the grounds that they had been funded with misappropriated customer money.[32][33]
The collapse triggered a period of intense self-examination inside the movement. A March 2023 Time report described how senior figures at the Centre for Effective Altruism had been warned about Bankman-Fried's conduct as far back as 2018 by people who had worked at Alameda Research, but those warnings had not stopped EA from publicly celebrating him as a model of "earning to give." Several leaders, including MacAskill, issued public apologies and reduced their public profiles for a time.[34]
EA has been the subject of sustained criticism from inside and outside the movement, from at least three distinct directions.
A first line concerns longtermism itself. The philosopher Émile P. Torres, who left the movement and became a vocal critic, has argued that strong longtermism uses the moral weight of vast hypothetical future populations to justify deprioritising urgent harms in the present. Together with the computer scientist Timnit Gebru, formerly of Google's Ethical AI team, Torres has used the acronym "TESCREAL" (transhumanism, extropianism, singularitarianism, cosmism, rationalism, effective altruism, longtermism) to describe what they characterise as a connected family of techno-utopian ideologies with intellectual roots they trace to twentieth-century eugenics. Torres and Gebru argue that this bundle of ideas has shaped the priorities of leading AI labs in ways that pose risks to marginalised groups.[35][36] Defenders within EA have rejected the framing as conflating distinct positions, but the term has gained traction in critical commentary.
A second cluster of criticisms targets EA's methodology. Critics inside the development-economics community have argued that cost-effectiveness rankings favour easily measurable interventions over harder-to-quantify but potentially more transformative work on systemic change. Within philanthropy more generally, writers including Ken Berger and Robert Penna of Charity Navigator described EA's habit of comparing causes against each other as "moralistic" and "elitist," a charge MacAskill addressed in print without backing down from the comparative project. Other critics have pointed out that the movement's expected-value reasoning, when applied to very low-probability events, can be exploited to justify almost any action by inflating the assumed payoff (so-called "Pascal's mugging" worries).[37]
A third strand focuses on EA's social composition and governance. The movement is heavily concentrated in Oxford, San Francisco, Berkeley, and London; its donor base is dominated by a small number of tech-fortunes, particularly Moskovitz and Tuna's Open Philanthropy; and its leadership has been criticised as Western-centric and demographically narrow. The FTX episode exposed the limits of the "earn to give" pipeline that 80,000 Hours had previously celebrated, with critics arguing that it created perverse incentives by treating very-high-earning careers (including in volatile or extractive industries) as ethically virtuous so long as the proceeds were eventually given away. There has also been ongoing reporting about cases of sexual misconduct and inadequate safeguards at EA gatherings.[34][38]
In the years since the FTX collapse, the movement has fragmented in observable ways. Some funders and researchers have pulled back from the EA brand while continuing to work on cause areas it championed, particularly AI safety. Others have publicly distanced themselves from longtermism and refocused on near-term global health and animal welfare programmes.[39]
Funding has concentrated. With the FTX-linked money gone and clawback proceedings continuing, the EA ecosystem is more dependent than ever on Open Philanthropy and a small group of secondary funders such as Jaan Tallinn's Survival and Flourishing Fund, Longview Philanthropy, and the Effective Altruism Funds. In 2025, Open Philanthropy renamed itself Coefficient Giving as part of an effort to reposition outside the EA brand while keeping its grantmaking priorities.[20]
At Oxford, the institutional landscape has narrowed. The Future of Humanity Institute was closed by the University of Oxford on April 16, 2024, after what FHI staff described as years of administrative friction with the Faculty of Philosophy. Several FHI alumni have continued similar work at successor projects, including the Forethought Foundation and independent research groups.[40]
The centre of gravity of EA-aligned giving has shifted further toward AI risk specifically, partly because of the rapid progress of frontier AI systems since 2022. Surveys of grant flows, EA Forum activity, and 80,000 Hours career recommendations all point in the same direction: more dollars and more people focused on AI alignment, governance, and information security, even as the broader "effective altruism" movement has lost some of its public visibility and brand strength.[18][20]