EA and Longtermist Wins and Losses

Quick Assessment

Key Links

Overview

Effective altruism (EA) is a philosophical and social movement that uses evidence and reason to identify the most effective ways to benefit others. Within EA, longtermism emphasizes positively influencing the long-term future — particularly by reducing existential risks from advanced AI, engineered pandemics, and other catastrophic threats.¹ Tracking the movement's successes and failures matters both for internal strategic learning and for external evaluation of whether EA's approach to philanthropy and cause prioritization delivers on its ambitious claims.

The movement has achieved substantial concrete wins: GiveWell raised $415 million and directed $397 million to cost-effective programs in metrics year 2024,² the Against Malaria Foundation — consistently one of GiveWell's top-recommended organizations — has protected over 667 million people,³ and CEAOrganizationCentre for Effective AltruismOxford-based organization that coordinates the effective altruism movement, running EA Global conferences, supporting local groups, and maintaining the EA Forum.Quality: 78/100 reports strong engagement growth in 2025 after post-FTX declines.⁴ On the longtermist side, hundreds of millions of dollars have flowed to existential risk research, AI safety work has gained mainstream policy traction, and California has signed bills directly regulating AI risk.⁵

However, the movement has also suffered significant setbacks. The collapse of FTX and fraud conviction of Sam Bankman-FriedPersonSam Bankman-FriedSam Bankman-Fried, convicted of fraud and sentenced to 25 years for misappropriating FTX customer funds, damaged both cryptocurrency markets and the effective altruism community, raising substantiv...Quality: 55/100 — who had publicly framed his financial activities in longtermist terms — severely damaged EA's credibility.⁶ The closure of the Future of Humanity InstituteOrganizationFuture of Humanity InstituteThe Future of Humanity Institute (2005-2024) was a pioneering Oxford research center that founded existential risk studies and AI alignment research, growing from 3 to ~50 researchers and receiving...Quality: 51/100 at Oxford in April 2024 — the field's founding research institution — marked the largest single institutional loss in the longtermist research ecosystem.⁷ Internal tensions between neartermist and longtermist factions persist, with some global health and animal welfare advocates reporting that the longtermist turn has made their work harder and worsened their reputations.⁸ Critics from multiple directions challenge longtermism's epistemological foundations, arguing that its calculations are highly sensitive to speculative assumptions about the far future.⁹

EA grantmaking reached its peak around 2022 and has contracted since, though the 80,000 Hours 2021 estimate of approximately $46 billion committed to EA causes (growing at approximately 37% annually since 2015) reflects the scale of capital that had been pledged rather than deployed.¹⁰ Open Philanthropy, the largest single EA-aligned funder, has directed more than $4 billion in total grants since its founding in 2017.¹¹

Cumulative Statistics

The table below aggregates self-reported impact figures from major EA-aligned organizations. Figures are drawn from organizational reports, EA Forum posts, and external analyses. Counterfactual attribution — the question of how much of this impact would have occurred without EA — is addressed in the notes column and in the dedicated attribution section below.

Counterfactual note on cumulative statistics: These figures represent gross reach and estimated impact, not net counterfactual impact. GiveWell's methodology attempts to estimate counterfactual value (would a donor have given elsewhere?), but for most metrics the counterfactual fraction is unknown. The EA Forum's 2025 funding data post notes that self-reported organizational metrics are difficult to compare across cause areas.¹²

Year-by-Year Highlights

This section provides a scannable chronological record of major events in the EA and longtermist movements. Type classifications: Win = outcome broadly favorable to EA/longtermist goals; Loss = setback or failure; Mixed = ambiguous or contested outcome. Attribution confidence reflects how directly EA action caused the outcome versus broader trends.

Notable Policy Record

This table covers major policy outcomes with documented EA involvement. "EA Attribution" rates how much of the outcome is plausibly attributable to EA advocacy versus independent forces, using a qualitative scale: High = EA was a leading driver; Moderate = EA was a significant but non-decisive contributor; Low = EA was a minor participant in a broader movement; Contested = attribution is disputed.

Policy record notes: The absence of biosecurity and pandemic preparedness policy entries reflects a genuine gap in the EA attribution record. EA-adjacent organizations including Johns Hopkins Center for Health SecurityOrganizationJohns Hopkins Center for Health SecurityThe Johns Hopkins Center for Health Security is a well-established biosecurity organization that has significantly influenced US policy on pandemic preparedness and biological threats, with recent ...Quality: 63/100 contributed to pandemic preparedness frameworks, and Open Philanthropy has funded biosecurity work, but tracking specific policy outcomes attributable to EA advocacy in this domain requires further documentation. Similarly, nuclear risk reduction receives Open Philanthropy funding (grants to organizations including the Federation of American Scientists and Ploughshares Fund) but no specific policy wins have been tracked here due to attribution difficulty.

History and Movement Trajectory

EA emerged from proto-communities in the late 2000s and early 2010s, coalescing around organizations like GiveWell (founded 2006), Giving What We CanOrganizationGiving What We CanComprehensive reference page on Giving What We Can covering its history, pledge structure, research approach, and criticisms; notes 10,000+ pledgers, \$340M+ donated historically, \$80M in 2024, an...Quality: 62/100 (founded 2009 at Oxford by Toby OrdPersonToby OrdComprehensive biographical profile of Toby Ord documenting his 10% AI extinction estimate and role founding effective altruism, with detailed tables on risk assessments, academic background, and in...Quality: 41/100), and 80,000 HoursOrganization80,000 Hours80,000 Hours is the largest EA career organization, reaching 10M+ readers and reporting 3,000+ significant career plan changes, with 80% of \$10M+ funding from Coefficient Giving. Since 2016 they'v...Quality: 45/100 (founded 2011).¹ CEA was founded in 2012 as an umbrella organization for GWWC and 80,000 Hours, and the term "effective altruism" was coined that year.¹⁹ By January 2013, GWWC had welcomed its 300th member; by 2022, more than 7,000 people from 95 countries had taken a pledge, collectively representing over $2.5 billion in pledged lifetime donations.¹⁹ An estimated $416 million was donated to effective charities identified by the movement in 2019, representing approximately 37% annual growth since 2015.¹

By 2021, an estimated $46 billion in funding had been committed to EA causes — growing approximately 37% per year since 2015, with much of the growth concentrated in 2020–2021 — with around $420 million deployed annually (roughly 1% of committed capital).¹⁰ In early 2020, roughly 60% of EA annual deployment flowed through Open Philanthropy, 20% from GiveWell, and 20% from other sources.¹⁰

The longtermist turn within EA accelerated in the late 2010s and early 2020s. Key markers of this shift include Open PhilanthropyOrganizationCoefficient GivingCoefficient Giving (formerly Open Philanthropy) has directed \$4B+ in grants since 2014, including \$336M to AI safety (~60% of external funding). The organization spent ~\$50M on AI safety in 2024...Quality: 55/100's first major AI safety grants in 2015; the growing influence of the Future of Humanity InstituteOrganizationFuture of Humanity InstituteThe Future of Humanity Institute (2005-2024) was a pioneering Oxford research center that founded existential risk studies and AI alignment research, growing from 3 to ~50 researchers and receiving...Quality: 51/100 at Oxford (founded 2005 by Nick Bostrom), which coined core longtermist terminology and hosted researchers including Anders Sandberg and Toby OrdPersonToby OrdComprehensive biographical profile of Toby Ord documenting his 10% AI extinction estimate and role founding effective altruism, with detailed tables on risk assessments, academic background, and in...Quality: 41/100; and Toby OrdPersonToby OrdComprehensive biographical profile of Toby Ord documenting his 10% AI extinction estimate and role founding effective altruism, with detailed tables on risk assessments, academic background, and in...Quality: 41/100's 2020 book The Precipice, which articulated a systematic framework for existential risk reduction. Will MacAskillPersonWill MacAskillComprehensive biographical reference on Will MacAskill covering his founding of EA organizations, academic work on moral uncertainty and longtermism, AGI preparedness advocacy, and controversies in...Quality: 60/100's 2022 book What We Owe the Future brought longtermism to mainstream audiences, achieving bestseller status.²⁰ That same year, the FTX Future FundOrganizationFTX Future FundThe FTX Future Fund was a major longtermist philanthropic initiative that distributed 132M USD in grants (including ~32M USD to AI safety) before collapsing with FTX's November 2022 bankruptcy, exp...Quality: 60/100 had committed more than $130 million in grants, mostly to longtermist causes, within four months of launching in February 2022.²⁰²¹

Open PhilanthropyOrganizationCoefficient GivingCoefficient Giving (formerly Open Philanthropy) has directed \$4B+ in grants since 2014, including \$336M to AI safety (~60% of external funding). The organization spent ~\$50M on AI safety in 2024...Quality: 55/100 recommended over $400 million in grants in 2021, including $300 million in support for GiveWell's recommendations (up from $100 million in 2020).¹³ Through 2022, roughly 70% of Open Philanthropy's total funding went toward areas in its Global Health and Wellbeing portfolio, and 30% went toward areas in its longtermist portfolio — making the "EA = longtermism" characterization inaccurate as a description of funding distribution.²² Following the FTX collapse in late 2022, Open Philanthropy paused most new longtermist funding commitments pending further review.²²

EA grantmaking has been on a downward trend since the 2022 peak, though GiveWell has maintained strong funding levels; in metrics year 2024 (February 2024 to January 2025), GiveWell raised $415 million and directed $397 million to cost-effective programs.² Open Philanthropy expected to recommend over $700 million in grants in 2023; as of June 2025, it had directed more than $4 billion in total grants since founding.¹¹ No single public source provides a comprehensive aggregate of all EA-aligned giving for 2023 or 2024 across all funders, which limits precise quantification of the post-2022 decline.

By 2025, CEAOrganizationCentre for Effective AltruismOxford-based organization that coordinates the effective altruism movement, running EA Global conferences, supporting local groups, and maintaining the EA Forum.Quality: 78/100 reported a turnaround, with 20–25% year-over-year engagement increases across all tiers — substantially exceeding its 7.5–10% growth target and reversing moderate declines in 2023–2024.⁴ This growth metric refers to year-over-year change from 2024 baselines, not recovery to 2022 peak levels; CEA's own strategy documents explicitly noted that engagement with its programs had declined year-over-year during 2023–2024 before beginning to recover.¹⁴

The FHI closed in April 2024 after the Faculty of Philosophy declined to renew staff contracts, following years of administrative friction and a hiring and fundraising freeze imposed in 2020.¹⁹ The specific reasons the Faculty imposed constraints on FHI have not been made fully public; factors cited in reporting include bureaucratic disputes over FHI's operating style, personnel issues, and controversy over a resurfaced 1996 email from Bostrom. Senior researchers attempted multiple rescue options — none succeeded.

Community size and demographics: As of January 2023, there were 362 known active EA groups worldwide — up from 233 in late 2020 (approximately 55% growth over two years), with active groups in 56 countries.²³ CEA supports EA groups across many countries and has held EAGx conferences internationally.²⁴ In 2024, 189 organizers from 34 countries participated in CEA's Organizer Support Programme.²⁵ The 2024 EA Survey (conducted by Rethink PrioritiesOrganizationRethink PrioritiesRethink Priorities is a research organization founded in 2018 that grew from 2 to ~130 people by 2022, conducting evidence-based analysis across animal welfare, global health, and AI governance. Th...Quality: 60/100) found that the USA (34.4%) and UK (13.5%) remained the countries with the largest proportions of EA respondents; 32% came from Europe (excluding the UK) and 20% from the rest of the world.²⁶ Local group membership rates varied substantially by country: US respondents (21.9%) and UK respondents (25.8%) had among the lowest rates of local group membership, while many countries showed 40–80% local group membership rates among survey respondents.

Documented Wins

Global Health and Development

EA's most verifiable successes lie in global health. GiveWell has directed $397 million to cost-effective programs in metrics year 2024 alone,² and the Against Malaria Foundation — consistently one of GiveWell's top-recommended organizations — has distributed enough insecticide-treated nets to protect over 667 million people, with an estimated 270,000 lives saved since its inception.³ GiveWell announced its largest single grant ever of $96.3 million to the Against Malaria Foundation.⁵

Other concrete outcomes include significant reductions in parasitic disease through deworming programs. Evidence Action's Deworm the World program launched in Tanzania in January 2026, targeting more than 10 million children currently at risk of or infected with soil-transmitted helminths and/or schistosomiasis, partnering with the government to provide deworming after external funding for the program was suspended in 2025.²⁷ Globally, Deworm the World reached 198 million children in India, Kenya, Nigeria, Pakistan, and Malawi in 2024 alone — a record number — and has delivered more than 2 billion treatments since its founding at an estimated cost of less than $0.50 per child per treatment. A GiveWell-recommended $4.4 million renewal grant secures programs in India, Kenya, Nigeria, and Pakistan until 2026.²⁸

Counterfactual note: GiveWell's own methodology acknowledges that counterfactual attribution is difficult — some fraction of donors to top charities would have donated similarly without GiveWell's recommendations, and some fraction of lives saved via malaria nets would have occurred via alternative programs. GiveWell applies a "leverage and funging adjustment" to cost-effectiveness estimates based on the probability that other counterfactual funding scenarios would occur in the absence of their charities' philanthropic spending, and applies this methodology to specific grantees.²⁹

No comprehensive independent peer-reviewed evaluations of GiveWell's aggregate impact claims have been published. GiveWell's methodology is documented publicly, and interventions like insecticide-treated nets and deworming have been studied via randomized controlled trials (e.g., through J-PAL-affiliated researchers), but external assessments of whether GiveWell-recommended programs achieve their stated impact in aggregate have not appeared in major peer-reviewed outlets.

Animal Welfare

Animal welfare has emerged as an increasingly prominent EA cause area. According to a 2025 EA Forum post tracking wins, 92% of corporate cage-free egg commitments with 2024 or earlier deadlines have been fulfilled.⁵ Lewis Bollard's appearance on the Dwarkesh Podcast raised over $2 million for effective animal welfare charities, estimated to help approximately 4 million animals.⁵ The 2024 EA Survey showed animal welfare joining the top tier of cause prioritization among community members.³⁰ Open Wing Alliance secured 141 new cage-free commitments in 2024 alone, per Open Philanthropy's 2024 progress report.¹⁷

Counterfactual note: Corporate cage-free commitments began accelerating in 2015–2016, partly driven by Humane Society campaigns and EU regulatory trends that predated significant EA funding of animal welfare campaigns. Open Wing Alliance (the main EA-funded vehicle) estimates it played a more decisive role in accelerating commitments in Southeast Asia and Latin America specifically, where independent consumer pressure was lower — a more plausible counterfactual claim than for US/EU markets.⁵ No published study or comparative campaign analysis has been identified that quantitatively estimates the geographic differential in EA-attributed impact versus broader societal trends.

Quality of fulfillment data: The 92% cage-free fulfillment figure tracks whether companies made formal policy transitions, not verified supply chain welfare outcomes at the farm level. The gap between corporate policy adoption and verified welfare improvement at the farm level remains an open measurement challenge.

AI Safety and Existential Risk Policy

The longtermist wing can point to growing policy influence. California and New York have signed bills directly related to AI risk regulation, following advocacy efforts supported by EA-aligned organizations.⁵ The Existential Risk Persuasion Tournament (XPT), which ran June–October 2022 (approximately 50 domain experts, approximately 50 superforecasters), found that approximately 42% of expert existential risk forecasters reported having attended an EA meetup, compared to 9% of superforecasters — providing the closest published proxy for EA community overlap with the x-risk researcher pool.³¹ This measures conference attendance, not career origin, and does not establish that EA was the primary cause of participants' entry into the field. AI safety has expanded significantly within AI/ML research communities, with historical ties to rationalist communities like LessWrongOrganizationLessWrongLessWrong is a rationality-focused community blog founded in 2009 that has influenced AI safety discourse, receiving \$5M+ in funding and serving as the origin point for ~31% of EA survey responden...Quality: 44/100 playing a role in building early researcher networks.³²

A 2022 survey of AI safety researchers at organizations including OpenAI, DeepMind, FAR AI, Open Philanthropy, Rethink Priorities, MIRI, Redwood, and GovAI found that all respondents had at least familiarity with EA and/or rationalist communities, with most being actively involved in at least one — and that Superintelligence and 80,000 Hours writing were each mentioned by three people as influential on their decision to work on AI safety.³² This is consistent with substantial EA influence on early researcher recruitment but does not establish the proportion of current AI safety researchers at major labs who entered the field through EA channels; no such systematic survey has been published.

Counterfactual note on AI safety wins: The central attribution challenge is temporal. EA began funding AI safety in 2015 when it was a niche concern; by 2023–2024, AI safety was mainstream following ChatGPT's release. Policy wins like Bletchley and Seoul were substantially driven by post-ChatGPT political momentum rather than EA-specific advocacy. The field would have grown substantially with the commercial AI trajectory regardless of EA funding. EA's contribution is more plausibly attributed to earlier field-building (2015–2021) than to post-2022 policy outcomes. A "commonly cited estimate" that EA meaningfully accelerated AI safety policy timelines has circulated in community discussions, but no named source, published model, or methodology has been identified — this claim should be treated as a qualitative community judgment rather than a quantitative estimate.

Community and Funding Infrastructure

Giving What We CanOrganizationGiving What We CanComprehensive reference page on Giving What We Can covering its history, pledge structure, research approach, and criticisms; notes 10,000+ pledgers, \$340M+ donated historically, \$80M in 2024, an...Quality: 62/100 reached 10,000 members taking the 10% pledge, with each pledge estimated to generate $15,000 in counterfactual donations to high-impact charities over a lifetime, per GWWC's Lifetime Giving Method.¹⁸ Partnership pledge drives at EA Global and EAGx events generated 203 new pledges during Q2–Q4 2024, estimated to produce $9.8 million in lifetime donations.¹⁴ This $9.8M figure is derived from GWWC's lifetime donation methodology applied to new pledges; the per-pledge lifetime estimate uses empirical GWWC retention data, external reference classes, and temporal discounts including inflation and global catastrophic risk uncertainty.¹⁸ Open PhilanthropyOrganizationCoefficient GivingCoefficient Giving (formerly Open Philanthropy) has directed \$4B+ in grants since 2014, including \$336M to AI safety (~60% of external funding). The organization spent ~\$50M on AI safety in 2024...Quality: 55/100 directed $87 million to GiveWell-recommended charities in 2024.¹⁷

Documented Losses and Setbacks

The FTX Collapse

The collapse of FTX and the fraud conviction of Sam Bankman-FriedPersonSam Bankman-FriedSam Bankman-Fried, convicted of fraud and sentenced to 25 years for misappropriating FTX customer funds, damaged both cryptocurrency markets and the effective altruism community, raising substantiv...Quality: 55/100, who had publicly framed his financial activities using longtermist reasoning about maximizing utility, damaged EA's credibility and raised concerns about ends-justify-the-means thinking within utilitarian frameworks.⁶ Total EA grantmaking has been on a downward trend since 2022.¹²

The FTX Future FundOrganizationFTX Future FundThe FTX Future Fund was a major longtermist philanthropic initiative that distributed 132M USD in grants (including ~32M USD to AI safety) before collapsing with FTX's November 2022 bankruptcy, exp...Quality: 60/100 launched in February 2022 and shut down in November 2022. During its existence, it made grants worth approximately $100 million and committed to $160 million in total grantee commitments as of September 2022.²¹ The fund's team resigned en masse, stating they were "unable to perform our work or process grants" and had "fundamental questions about the legitimacy and integrity of the business operations" funding the Future Fund.³³ All unfulfilled commitments were voided when FTX entered bankruptcy.

Among affected organizations: SBF's biggest grants went to pandemic prevention and EA institutions including CEAOrganizationCentre for Effective AltruismOxford-based organization that coordinates the effective altruism movement, running EA Global conferences, supporting local groups, and maintaining the EA Forum.Quality: 78/100, the Long Term Future Fund, Lightcone Infrastructure, The Atlas Fellowship, and Constellation. University-affiliated research initiatives received more than $13 million in total; twenty academics at institutions including Cornell, Princeton, Brown, and Cambridge received individual grants exceeding $100,000 each.³⁴ Other affected grantees included HelixNano and Our World in Data.³⁵ The full organizational landscape — which entities closed versus found alternative funding — has not been comprehensively documented in any single public source.

CEA's own reports and the 2025 EA Forum funding data post document the subsequent multi-year decline in forum engagement, EAG attendance, EA Funds donations, and virtual programming participation that followed the collapse.¹²¹⁴

The Closure of the Future of Humanity Institute

The closure of the Future of Humanity InstituteOrganizationFuture of Humanity InstituteThe Future of Humanity Institute (2005-2024) was a pioneering Oxford research center that founded existential risk studies and AI alignment research, growing from 3 to ~50 researchers and receiving...Quality: 51/100 (FHI) at Oxford on April 16, 2024 represents the largest single institutional loss in the longtermist research ecosystem.⁷ Founded in 2005 as part of the Faculty of Philosophy and the Oxford Martin School, FHI had Nick Bostrom as director and staff including Anders Sandberg and Toby OrdPersonToby OrdComprehensive biographical profile of Toby Ord documenting his 10% AI extinction estimate and role founding effective altruism, with detailed tables on risk assessments, academic background, and in...Quality: 41/100. The institute coined much of the field's foundational terminology — "existential risk," "existential hope," "information hazard," "unilateralist's curse" — and hosted research that shaped global biosecurity and AI safety policy debates.

Beginning in 2020, the Faculty of Philosophy imposed a freeze on FHI fundraising and hiring. In late 2023, the Faculty announced it would not renew contracts of remaining FHI staff. Senior researchers attempted multiple rescue options — leadership restructuring, joining a college, spinning out of the university entirely, and an interdepartmental transfer to Physics — none succeeded. The University of Oxford's closing statement cited "increasing administrative headwinds within the Faculty of Philosophy."⁷ The specific reasons the Faculty imposed constraints on FHI have not been made fully public; factors cited in reporting include bureaucratic disputes over FHI's fast-moving, externally-networked operating style, personnel issues, and the controversy over a resurfaced 1996 email from Bostrom.³⁶

FHI's final report, authored by Anders Sandberg, stated that "FHI's mission has replicated and spread and diversified" into dozens of organizations and thousands of individuals — noting successor institutions including the UK AI Safety InstituteOrganizationUK AI Safety InstituteThe UK AI Safety Institute (renamed AI Security Institute in Feb 2025) operates with ~30 technical staff and 50M GBP annual budget, conducting frontier model evaluations using its open-source Inspe...Quality: 52/100, the Global Priorities Institute (also at Oxford), and the Institute for Ethics in AI.⁷ Oxford retains other EA-adjacent research institutions, limiting but not eliminating the institutional loss.

Internal Tensions

The shift toward longtermism has generated friction within the EA community. Some EA advocates focused on global health and animal welfare have reported that longtermism and x-risk work has made their lives harder, worsened their reputations, and occupied valued community niches.⁸ The 2024 EA Survey revealed diverging priorities: highly engaged respondents rated AI risks, animal welfare, and EA movement building more highly, while less engaged members emphasized climate change, global health, and poverty.³⁰ An EA Forum post on the relationship between the EA community and AI safety noted that AI safety dominance has alienated global health and development EAs, creating perceptions of exclusion at events, and that community members sometimes perceive EA as an AI/longtermism-only movement — a perception that risks talent loss in other cause areas.³⁷

Funding Decline

While GiveWell has maintained relatively stable and growing funding from non-Open Philanthropy donors (approximately $415M raised in 2024), the broader EA funding landscape contracted since the FTX collapse.¹²² Open Philanthropy's available assets fell by roughly half over the course of 2022, though they had since recovered about half of the total losses by 2023; this led Open Philanthropy to raise the cost-effectiveness bar for Global Health and Wellbeing grants by roughly a factor of two.³⁸ Open Philanthropy does not publish audited balance sheets publicly; the "recovered half" figure appears in a Coefficient Giving summary of Open Philanthropy's own communications rather than in an independently audited source.³⁸

MIRIOrganizationMachine Intelligence Research InstituteComprehensive organizational history documenting MIRI's trajectory from pioneering AI safety research (2000-2020) to policy advocacy after acknowledging research failure, with detailed financial da...Quality: 50/100's strategic pivot away from technical alignment research (announced January 2024) was one visible indicator of funding pressure and strategic reorientation at longtermist-focused organizations. MIRI's annual spending ranged from $5.4M–$7.7M during 2019–2023, with a high in 2020 and a low in 2022; projected 2024 spending was $5.6M. MIRI's own framing of the change was a strategic scaling back rather than a traditional layoff.¹⁵

Policy Setbacks

California SB 1047's veto by Governor Newsom in September 2024 was the most visible EA-supported AI safety policy loss. The bill would have imposed safety requirements on large AI model developers. Anthropic's position on the bill evolved: the company initially stated it did not support SB 1047 in its original form in July 2024, proposed specific amendments (including removing pre-harm civil penalty authority for the attorney general and shifting the legal standard from "reasonable assurance" to "reasonable care"), saw those amendments partially adopted, then offered cautious support for the amended version in August 2024 — while noting "there are still some aspects of the bill which seem concerning or ambiguous."¹⁶ At least 113 current and former AI company employees, including some at Anthropic, signed a letter to Newsom supporting the bill before his veto.³⁹ Newsom vetoed the bill citing economic competitiveness concerns.

Counterfactual Attribution Analysis

Evaluating EA's impact requires distinguishing between outcomes EA caused and outcomes EA coincided with. The following framework summarizes attribution confidence across cause areas.

High-confidence EA attribution:

• GiveWell's redirection of individual donor funds to specific charities: GiveWell's recommendation methodology is the proximate cause of most directed funds; without GiveWell, many donors would have given to lower-impact charities or not at all. GiveWell's counterfactual methodology is documented publicly, though specific estimates vary by grantee and funding round.²⁹
• Charity Entrepreneurship incubatees: organizations including LEEP that would not exist without CE's program and which have documented policy wins in multiple countries.
• Giving What We CanOrganizationGiving What We CanComprehensive reference page on Giving What We Can covering its history, pledge structure, research approach, and criticisms; notes 10,000+ pledgers, \$340M+ donated historically, \$80M in 2024, an...Quality: 62/100 pledges: the pledge mechanism directly influenced donation behavior for 10,000+ members, with methodology documented in GWWC's 2023–2024 impact evaluation.¹⁸

Moderate-confidence EA attribution:

• AI safety researcher pipeline (pre-2022): EA significantly influenced career entry for many researchers now at AI safety organizations. Evidence includes the XPT data (42% of expert x-risk forecasters attended EA meetups)³¹ and qualitative surveys showing universal familiarity with EA among early AI safety researchers.[^49] EA's marginal contribution to the field's current size is harder to isolate given the post-2022 commercial AI boom.
• Corporate animal welfare commitments in emerging markets: EA-funded campaigns have a more plausible counterfactual impact in markets with less independent consumer pressure (Southeast Asia, Latin America) than in US/EU markets where broader societal trends were also driving change. No published comparative analysis quantifying this geographic differential has been identified.
• Open Philanthropy's R21 co-funding: contributory but not decisive; Wellcome Trust and Serum Institute of India were primary funders.

Low-confidence EA attribution:

• Multilateral AI safety declarations (Bletchley, Seoul): post-ChatGPT political momentum was the primary driver; EA-aligned researchers contributed technical input but no named individuals or government acknowledgments documenting EA's specific causal role have been published.
• EU AI ActPolicyEU AI ActComprehensive overview of the EU AI Act's risk-based regulatory framework, particularly its two-tier approach to foundation models that distinguishes between standard and systemic risk AI systems. ...Quality: 55/100 provisions: primarily driven by EU parliamentary process and civil society groups; EA input was marginal.
• US AI safety mainstream adoption post-2022: the field scaled with AI capabilities, not primarily with EA funding.

Active attribution disputes:

• Whether longtermist cause prioritization crowded out neartermist funding. Global health advocates argue yes; longtermists argue the donor pools are largely separate. Open Phil's own published allocation data (70% global health / 30% longtermist through 2022) partially addresses this for financial capital, but the question of talent and attention allocation is harder to resolve.²²
• Whether EA's early AI safety focus shaped OpenAIOrganizationOpenAIComprehensive organizational profile of OpenAI documenting evolution from 2015 non-profit to Public Benefit Corporation, with detailed analysis of governance crisis, 2024-2025 ownership restructuri...Quality: 62/100's and AnthropicOrganizationAnthropicComprehensive reference page on Anthropic covering financials (\$380B valuation, \$19B ARR), safety research (Constitutional AI, mechanistic interpretability, model welfare), governance (LTBT struc...Quality: 74/100's safety cultures, versus those organizations developing safety norms independently.

Criticisms and Limitations

Longtermist Modeling Vulnerabilities

Research published on the EA Forum has identified critical vulnerabilities in longtermist calculations: small changes in baseline existential risk assumptions can dominate tenfold differences in period risk reduction estimates, and the value of existential risk mitigation remains highly sensitive to estimated future risk levels.⁹ The "time of perils hypothesis" — used to defend both the claim that existential risks are high and that reducing them is especially valuable — requires speculative claims about how risks will diminish after a certain period.⁹

Philosopher Steven Pinker has critiqued longtermism for potentially prioritizing any scenario, no matter how improbable, as long as it can be framed as having arbitrarily large effects far in the future.⁴⁰ A related concern is that longtermism's predictive confidence should be very low given the exponential branching of possible futures.⁴⁰

Systemic and Philosophical Objections

Critics argue that EA focuses on symptoms via charity rather than addressing root causes like institutional reform, debt, and power inequality. Philosopher Amia Srinivasan has argued that "Effective Altruism doesn't try to understand how power works, except to better align itself with it. In this sense it leaves everything just as it is."⁴¹ In academic formulations, Srinivasan contends that by insisting on quantifying the value of actions in terms of potential effects and probabilities of success, EA "effectively endorses the prevailing global capitalist institutional order — precisely the order that must be changed to address global poverty meaningfully."⁴² Nathan Robinson published a widely discussed critique in Current Affairs in 2023 making related institutional arguments.⁴³ Brian Berkey's peer-reviewed "Institutional Critique of Effective Altruism" (Utilitas, Cambridge, 2017) formalizes these claims in academic philosophy.⁴²

EA proponents have responded to institutional critiques by noting that 80,000 Hours' stated priorities include "mitigating great power conflict," "global governance," and "space governance" — pointing to engagement with structural questions that go beyond individual charity.⁴⁴

Evidence Standards and Epistemic Concerns

GiveWell itself has acknowledged that seeking strong evidence and a straightforward, documented case for impact can be in tension with maximizing impact — reflecting a deep epistemic split within EA between evidence-based giving and longtermism's higher-risk, higher-reward approach.⁶ Jacob Steinhardt has argued that EA has a history of making overconfident claims with insufficient research, citing Peter Singer's 2009 claim that a life could be saved for $200 — a figure substantially revised by 2011.⁴⁵

Ben Kuhn has argued that EA exhibits "epistemic inertia," where community consensus becomes less responsive to new evidence and arguments, partly because group norms fail to adequately guard against motivated cognition.⁴⁶

Current Strategic Direction

CEAOrganizationCentre for Effective AltruismOxford-based organization that coordinates the effective altruism movement, running EA Global conferences, supporting local groups, and maintaining the EA Forum.Quality: 78/100 is pursuing what it describes as a growth-focused stewardship strategy for 2025–2026, aiming to build sustainable momentum. The organization acknowledges that meaningful results may not materialize until the end of 2026, as it requires building new infrastructure and iterating on strategy.¹⁴

Will MacAskillPersonWill MacAskillComprehensive biographical reference on Will MacAskill covering his founding of EA organizations, academic work on moral uncertainty and longtermism, AGI preparedness advocacy, and controversies in...Quality: 60/100 has argued in a widely discussed October 2025 post that EA should expand beyond traditional cause areas to address the transition to a post-AGI society, rejecting both "legacy movement" and "refocus on classic causes" framings. MacAskill specified that in terms of what people do, perhaps 15% of EA effort (scaling to 30% or more over time) should be primarily working on cause areas beyond the "classic" four (AI safety, biorisk, animal welfare, global health) — including AI welfare, AI character, AI persuasion and epistemic disruption, human power concentration, and space governance. He further proposed that in terms of curriculum content, perhaps 30% or more should cover non-classic cause areas.⁴⁷ These figures are from MacAskill's own October 2025 post, based on a memo written for the Meta Coordination Forum, and are not derived from an independent survey or community mandate.

GiveWell continues recommending charities in mental health, malnutrition, and lead exposure while broadening its work beyond charity evaluation to identify new grantees and explore underexplored cause areas.⁴⁸ Longtermist projects represented less than one-third of total EA funding as of August 2022, with the majority of EA work falling outside longtermist frameworks.⁴⁹

Key Uncertainties

• Post-FTX funding trajectory: Whether EA grantmaking stabilizes at current levels, continues declining, or recovers remains unclear. Open PhilanthropyOrganizationCoefficient GivingCoefficient Giving (formerly Open Philanthropy) has directed \$4B+ in grants since 2014, including \$336M to AI safety (~60% of external funding). The organization spent ~\$50M on AI safety in 2024...Quality: 55/100 has begun efforts to diversify its donor base beyond Good Ventures.¹⁷ No comprehensive public breakdown of total EA-aligned giving for 2023 or 2024 across all funders has been published, making quantification of the post-2022 decline difficult.
• AI safety policy impact: While California and New York have passed AI safety legislation, the actual effect of these bills on reducing catastrophic AI risk is untested and contested.
• Longtermist epistemology: The sensitivity of longtermist calculations to baseline assumptions about existential risk levels means that the case for longtermist interventions could strengthen or weaken substantially with new evidence.⁹
• Community cohesion: Whether EA can maintain unity across its neartermist and longtermist wings, or whether increasing specialization leads to effective fragmentation, remains an open question.⁸³⁰
• Reputational recovery: The extent to which the FTX scandal permanently affected EA's ability to attract talent and funding versus representing a temporary setback is still unfolding.⁶
• AI safety attribution: As AI safety becomes mainstream post-ChatGPT, EA's distinctive comparative advantage in the field may diminish even as absolute activity increases — a strategic question the community has not fully resolved.³⁷
• Cage-free commitment quality: The 92% fulfillment figure for commitments with 2024 deadlines tracks whether companies made formal policy transitions, not verified supply chain welfare outcomes at the farm level. The gap between corporate policy adoption and verified welfare improvement remains an open measurement challenge.
• FHI institutional legacy: Whether the closure of FHI represents a net setback for longtermist research — or whether successor organizations (UK AISI, Global Priorities Institute, etc.) have absorbed and expanded FHI's research agenda — is contested within the community.⁷
• Mental health as EA cause area: GiveWell recommends StrongMinds as a top charity focused on mental health, but no aggregate impact metrics for mental health interventions comparable to the lives-saved estimates for malaria or deworming programs have been published in this page's tracked data.

Sources