FTX Collapse: Lessons for EA Funding Resilience

Quick Assessment

Overview

On November 11, 2022, FTX—at the time valued at $32B—filed for bankruptcy, eliminating what had been the EA ecosystem's second-largest funding source. 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 had positioned himself as EA's most prominent public donor, pledging billions through the FTX Foundation and its 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 grantmaking arm.

The collapse unfolded over approximately 72 hours. The EA ecosystem's immediate losses included approximately $160M in committed Future Fund grants that were not disbursed, $26.8M returned by Effective Ventures to FTX's bankruptcy estate under clawback proceedings, and the elimination of what had been projected to become one of the largest EA funding sources in history—the Future Fund had originally projected deploying up to $1B in 2022.⁵

Three years later, the EA community has partially recovered financially—Coefficient GivingOrganizationCoefficient 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 explicitly expanded grantmaking to fill some of the gap⁶—but structural questions about funding concentration persist. The FTX episode is directly relevant to the current period: an estimated $27–76B in risk-adjusted EA-aligned capital is now tied to AnthropicOrganizationAnthropicComprehensive reference page on Anthropic covering financials (\$380B valuation, \$14B ARR), safety research (Constitutional AI, mechanistic interpretability, model welfare), governance (LTBT struc...Quality: 74/100.

Chronology of Key Events

The FTX Future Fund: Background

The FTX Future Fund officially launched in February 2022 as a project of the FTX Foundation, with the stated mission of "making grants and investments to ambitious projects in order to improve humanity's long-term prospects."⁷ At launch, the team included Nick BecksteadPersonNick BecksteadNick Beckstead is a philosopher and EA/longtermism figure whose 2013 dissertation formalized longtermist ethics; this article covers his career arc from academic philosopher to Coefficient Giving (...Quality: 60/100 (CEO), Leopold AschenbrennerPersonLeopold AschenbrennerComprehensive biographical profile of Leopold Aschenbrenner, covering his trajectory from Columbia valedictorian to OpenAI researcher to \$1.5B hedge fund founder, with detailed documentation of hi...Quality: 61/100, 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, and Ketan Ramakrishnan. Beckstead had joined the FTX Foundation as CEO in November 2021, having previously spent seven years as an early employee at Coefficient Giving and as a Research Fellow at the Future of Humanity Institute at Oxford; he also helped launch the Centre for Effective Altruism.¹⁵

At launch, the Fund planned to distribute at least $100M in its first year, with a stated ambition of potentially deploying up to $1B.⁷ It operated for approximately nine months before FTX's collapse in November 2022.

Post-Collapse: Future Fund Leadership

The Future Fund's leadership largely avoided public statements after the November 2022 resignation letter, with media attempts to contact Beckstead, Aschenbrenner, Ramakrishnan, and MacAskill via LinkedIn, Twitter, and email in early 2023 going unanswered.¹⁰

Beckstead subsequently stepped down from the boards of Effective Ventures UK and Effective Ventures US on August 23, 2023. Since November 2022, he had been recused from all board matters related to the FTX collapse, and over time that recusal made it difficult for him to contribute meaningfully to EV's governance.¹⁰ He subsequently co-founded and serves as CEO of the Secure AI Project, focused on developing and advocating for policies to reduce risks from advanced AI.¹⁶

The FTX-EA Funding Relationship

Scale and Scope

Before the collapse, FTX-linked funding touched nearly every corner of the EA ecosystem:

The committed/disbursed distinction: The $160M figure on the Future Fund website referred to "grantee commitments," not confirmed disbursements.¹ As of June 2022, $132M had been reported as awarded across 262 grants and investments, though how much had actually been physically transferred to grantees before the collapse was not publicly established.⁸ The Future Fund team's resignation letter stated they were "devastated that it looks like there are many committed grants that the Future Fund will be unable to honor."⁹ Some grantees who had received funds before the collapse faced the ethical dilemma of using those funds or returning them, fearing clawback under US bankruptcy law.¹⁷ The number of outstanding unfulfilled commitments at the time of collapse (as distinct from the 262 grants and investments reported as of June 2022) has not been publicly documented.

Concentration Risk in Context

In 2021–2022, the EA funding landscape had two dominant sources:

1. Dustin MoskovitzPersonDustin Moskovitz (AI Safety Funder)Dustin Moskovitz and Cari Tuna have given \$4B+ since 2011, with ~\$336M (12% of total) directed to AI safety through Coefficient Giving, making them the largest individual AI safety funders global...Quality: 49/100 (via Coefficient GivingOrganizationCoefficient 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 / Good Ventures): approximately $300–500M/year
2. 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 (via FTX Foundation / Future Fund): approximately $160M committed in roughly nine months of operation (February–November 2022)

One EA Forum analysis estimated that FTX and Alameda co-founders' wealth made up approximately 35% of all funding available to EA before the collapse—though the methodology for that estimate is not described in the forum post, which is not a peer-reviewed source.²⁰ The claim that the two dominant sources together controlled 80–90% of EA-directed funding is sometimes cited in community discussions; this figure is not independently sourced and appears to derive from the same forum post combined with unstated assumptions about Coefficient Giving's share. Readers should treat both figures as approximate community estimates rather than systematically derived data.

What is better documented is the directional impact: subsequent EA funding tracker data confirm a sustained decline across all major categories after 2022. Comparing 2024 to 2022 levels, EA grantmaking declined approximately 9% in global health and development, approximately 13% in longtermism/x-risk, approximately 23% in animal welfare, and approximately 21% in meta/EA infrastructure.²¹

FTX Bankruptcy Scale

At the time of filing, FTX had more than one million creditors, according to an early bankruptcy filing. The largest single creditor was owed $226M; the top 50 creditors collectively held approximately $3.1B in unsecured claims.²² FTX ultimately recovered between $14.7B and $16.5B in assets for distribution, and a bankruptcy plan approved by the court provided that 98% of creditors would receive 119% of what they were owed based on bankruptcy-filing-date values.²³ This full-creditor-repayment outcome is relevant context for understanding why clawback pressure on EA organizations ultimately dissipated.

Factors Contributing to Losses

This section describes failure modes identified in retrospective analyses by EA organizations, grantees, and observers. The characterizations reflect documented outcomes, not assessments of intent.

1. Spending Against Unreceived Commitments

The most operationally disruptive pattern was organizations hiring staff and launching programs based on Future Fund commitments before funds were received. When commitments evaporated:

• Organizations had payroll obligations they could not meet
• Multi-year projects were abandoned mid-stream
• Staff who relocated for EA-adjacent jobs found themselves without positions within months
• At least one PhD candidate at Cambridge had to leave their program when committed funding never arrived¹⁷

EA Funds' Long-Term Future Fund and Infrastructure Fund reported estimated funding gaps of approximately $450K/month and $550K/month respectively over the six months following FTX's collapse.²⁴

Relevance for Anthropic-linked capital: Organizations that budget against expected future Anthropic liquidity events before capital is received face analogous structural risk. Even DAF-committed capital can be redirected to non-EA causes at the DAF advisor's discretion.

2. Insufficient Scrutiny of Funding Sources

Post-collapse analyses identified several characteristics of FTX's business that received limited attention within EA before the collapse:

• FTX's trading volume and profitability claims relative to its market position
• SBF's "earn-to-give" framing and potential incentive misalignment
• Alameda Research's relationship with FTX and corporate governance questions
• Community members raised concerns that did not receive significant public attention at the time

Relevance for Anthropic-linked capital: The relevant parallel is concentration risk, not fraud risk. AnthropicOrganizationAnthropicComprehensive reference page on Anthropic covering financials (\$380B valuation, \$14B ARR), safety research (Constitutional AI, mechanistic interpretability, model welfare), governance (LTBT struc...Quality: 74/100 is a legitimate business with disclosed governance structures. However, the structural question of dependence on a single entity's outcomes applies regardless of that entity's integrity.

3. Reputational Association

The FTX collapse produced association between "effective altruism" and "crypto fraud" in mainstream coverage. Articles in major outlets—including the New York Times, the Guardian, the Washington Post, New York Magazine, and The Economist—characterized EA in critical terms during late 2022 and 2023, using phrases such as "morally bankrupt," "ineffective altruism," and "defective altruism."²⁵ Documented effects include increased difficulty in some talent recruitment contexts and organizational time redirected to reputational management.

One documented example of institutional distancing: CSET (Center for Security and Emerging Technology at Georgetown University), which received more than $100M from Coefficient Giving, had its executive director praise the EA movement after SBF's fall in December 2022—but by November 2023 stated publicly that "CSET is not an EA organization."²⁶ Anthropic CEO Dario AmodeiPersonDario AmodeiComprehensive biographical profile of Anthropic CEO Dario Amodei documenting his competitive safety development philosophy, 10-25% catastrophic risk estimate, 2026-2030 AGI timeline, and Constituti...Quality: 41/100 publicly distanced himself from the EA movement after the collapse, expressing concern that EA's institutional culture can be insular and that its emphasis on abstract long-term reasoning can come at the expense of practical judgment.²⁷

Whether and to what degree these effects have persisted or reversed is contested. EA Funds experienced its first month-over-month donor decline in December 2022 after unbroken growth since 2017; donor participation declined approximately 10–20% in 2023; and Giving What We Can pledge data showed approximately a 70% decline in new monthly pledges in 2023 versus 2022.²⁸ Some observers note continued EA organizational activity as evidence of resilience. The Hollywood Reporter described EA's "Post-SBF Comeback" in mid-2025.² No systematic survey of public perceptions before and after the collapse, or content analysis of media coverage, has been published.

Relevance for Anthropic-linked capital: If Anthropic experienced a major safety failure or public controversy, EA organizations funded by Anthropic-linked capital would face reputational questions regardless of their actual relationship to the event—the same mechanism that operated in 2022. "EA = Anthropic insiders" appears as a framing in some critical coverage (e.g., critiques of EA's relationship with AI labs²⁹), though no media analysis or survey data documenting the prevalence of this framing in public discourse has been published.

4. Legal and Financial Exposure

The FTX bankruptcy created unexpected legal liabilities for recipient organizations:

• Clawback proceedings: FTX's estate pursued recovery under fraudulent transfer laws. Effective Ventures (the umbrella over 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, 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, and 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) returned $26,786,503—equal to 100% of funds received from FTX and the FTX Foundation in 2022.³ An independent investigation by Mintz found no evidence that anyone at EV knew about FTX's alleged fraudulent conduct.³
• Other organizations: Almost all other potential clawbacks were described by EV as "relatively small potatoes," many below the public filing threshold.³ ARCOrganizationAlignment Research CenterComprehensive reference page on ARC (Alignment Research Center), covering its evolution from a dual theory/evals organization to ARC Theory (3 permanent researchers) plus the METR spin-out (Decembe...Quality: 57/100 voluntarily returned $1.25M without being compelled.¹⁸ Center for Applied RationalityOrganizationCenter for Applied RationalityBerkeley nonprofit founded 2012 teaching applied rationality through workshops (\$3,900 for 4.5 days), trained 1,300+ alumni reporting 9.2/10 satisfaction and 0.17σ life satisfaction increase at 1-...Quality: 62/100 and its affiliated entity Lightcone Infrastructure were named as defendants in an adversarial proceeding; FTX sought to recover $4.9M, comprising a $2M grant to CFAR and additional amounts to Lightcone Infrastructure, which owns the Lightcone Rose Garden conference venue. Court filings dated May 17, 2024 summoned CFAR and Lightcone to appear in court; CFAR had reportedly ignored FTX trustees for months in 2023, only responding after a discovery motion was filed in October 2023.¹⁹ A separate EA-affiliated nonprofit filed to dismiss a $5M clawback lawsuit in July 2024;³⁰ the outcome of this motion—whether dismissed, settled, or otherwise resolved—has not been reported in publicly available sources as of this writing.
• Statute of limitations: The deadline for FTX to file new clawback claims expired in November 2024. The bankruptcy estate intensified filings in early November 2024, two years after the petition date.³¹ Grant recipients outside the United States may still face exposure under longer limitation periods in their jurisdictions (e.g., six years under New Zealand law).¹² No publicly available reporting documents outcomes of non-US EA organization clawback claims after November 2024.
• Full creditor repayment: FTX's bankruptcy estate reported it expected to pay creditors in full, which raised questions about whether outstanding clawback proceedings against grantees would remain necessary.³² The estate still filed suits before the statute of limitations expired. The final resolution of those suits against EA organizations beyond EV—including the CFAR proceeding and the unnamed nonprofit's $5M case—is not publicly documented as of this writing.

SBF's conviction on November 2, 2023 and sentencing to 25 years on March 28, 2024 legally confirmed fraud, which strengthened the legal basis for fraudulent conveyance clawback claims (rather than mere preference claims) against recipients—though no specific reporting confirms the sentencing directly altered pending EA organization settlements.³³

Relevance for Anthropic-linked capital: Large concentrated donations can create legal complexity even when the donor acts in good faith. Tax complications, insider trading concerns around IPO timing, and SEC scrutiny of coordinated selling are potential risks analyzed at EA Shareholder Diversification from AnthropicE851This article documents Jeffrey Epstein's financial relationships with AI researchers including Ben Goertzel, Marvin Minsky, Joscha Bach, Seth Lloyd, Nick Bostrom, and MIRI, drawing on DOJ files and...Quality: 60/100.

FTX vs. Anthropic: Parallels and Differences

This section compares structural features of the two funding concentration scenarios. Given the documented failure modes above, the comparison focuses on concentration risk—not on fraud risk. Anthropic is a legitimate technology company with disclosed governance; the relevant parallel is single-source dependence, not the underlying cause of the 2022 collapse.

Parallels

Critical Differences

What Is Structurally Better

1. DAFs provide legal commitment to charity. Capital in donor-advised funds is legally required to go to 501(c)(3) organizations (though not necessarily EA causes). FTX commitments had no such legal backing.[^34]

2. Diversified holders. Unlike FTX (one person), Anthropic capital is distributed across founders, employees, and investors. Different holders may follow through on EA-aligned giving regardless of others' choices.

3. Gradual liquidity. An IPO event provides a more gradual, foreseeable liquidity timeline rather than overnight collapse. Organizations can plan around it.

4. Legitimate business with real revenue. Anthropic generates substantial revenue from real products. A valuation decline is a distinct risk category from the total-loss scenario that characterized FTX.

What Is Structurally More Complex

1. Far larger magnitude. The $27–76B risk-adjusted estimate is 170–500x larger than the FTX funding gap.

2. Longer timeline may reduce urgency. FTX's sudden collapse forced rapid organizational adaptation. A multi-year liquidity timeline may reduce the perceived urgency of resilience planning.

3. Potential correlation with AI outcomes. Anthropic's value is tied to AI development trajectories. It has been hypothesized that scenarios where AI safety funding demand is highest could coincide with increased competitive pressure on Anthropic's market position—but this relationship is not established empirically, and the direction of any correlation is uncertain. This should be understood as a hypothesis, not a documented structural risk.

4. Value alignment uncertainty over time. As Anthropic (Funder)AnalysisAnthropic (Funder)Comprehensive model of EA-aligned philanthropic capital at Anthropic. At \$380B valuation (Series G, Feb 2026, \$30B raised): \$27-76B risk-adjusted EA capital expected. Total funding raised exceed...Quality: 65/100 documents, the proportion of Anthropic employees with EA-affiliated giving commitments is an open question across successive hiring cohorts. Anthropic offered employees the option to pledge up to 50% of their equity to nonprofits (with a 3:1 company match) for most of its history, though post-2024 hires have a 1:1 matching structure for pledges up to 25%.³⁴ Some early employees—including Dario AmodeiPersonDario AmodeiComprehensive biographical profile of Anthropic CEO Dario Amodei documenting his competitive safety development philosophy, 10-25% catastrophic risk estimate, 2026-2030 AGI timeline, and Constituti...Quality: 41/100, who was the 43rd signatory of the Giving What We Can pledge, and Amanda Askell, the 67th signatory—have documented EA-affiliated giving commitments.³⁵ No systematic public data exists on pledge rates across successive hiring cohorts, and the company has made a communications decision to distance itself from the EA brand.³⁵ A multi-year liquidity timeline means an increasing share of capital will be controlled by employees hired in later cohorts.

Ecosystem Response to FTX

Coefficient Giving's Gap-Filling

Coefficient GivingOrganizationCoefficient 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 explicitly responded to FTX's collapse by opening applications to affected grantees and lifting a pause on new funding commitments to its Global Catastrophic Risks portfolio in early 2023.⁶ Coefficient Giving sought specifically to fill gaps in biosecurity, AI risk, and longtermist EA community building left by the Future Fund's collapse.³⁶

Coefficient Giving's own asset base fluctuated significantly during this period: available assets fell by roughly half due to Moskovitz stock price decline, then partially recovered.³⁶ Coefficient Giving expected to recommend over $700M in grants in 2023.

However, Coefficient Giving did not fully replace FTX's disbursements to GiveWell: Coefficient Giving gave $100M to GiveWell in 2023, compared to $300M in 2021 and $350M in 2022. GiveWell directed $216M in total funds in 2023, down 51% from $439M in 2022. GiveWell attributed the drop primarily to the reduction in Coefficient Giving contributions; donations from non-Coefficient Giving sources remained relatively steady at approximately $255M (2023) versus $253M (2022).³⁷ Whether GiveWell independently adjusted its grantmaking targets in response to changing top-charity funding needs, and whether other funders increased or decreased their own commitments during this period, are not fully addressed in GiveWell's public reporting.

New Grantmaking Bodies

Five new independent grantmaking bodies started in 2023, nearly all focused on longtermism or AI safety, reflecting the ecosystem's response to the funding gap.²⁴ This partially diversified the grantmaking infrastructure, though the underlying capital sources for many of these bodies remain concentrated.

Organizational Resilience Measures

Several EA organizations documented specific resilience measures adopted after FTX:

• 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 explicitly targeted maintaining 12 months of operational reserves by end of 2024, and began excluding Coefficient Giving support from funding plans to assess true diversification status.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 maintained $16M in end-of-year reserves (as of December 2024), projected to fund activities for just over two years—consistent with their stated minimum of at least two years of reserves. MIRI's annual expenses ranged from $5.4M to $7.7M in 2019–2023, with projected 2024 spend of $5.6M.³⁹
• EA Infrastructure Fund (EAIF) documented efforts toward increased funder diversity, with Coefficient Giving employees stepping down from EAIF leadership positions and the fund planning to rely more on individual donors.⁴⁰
• GiveWell secured a $300M multi-year commitment from Coefficient Giving over 2023–2025 as a structural resilience measure, while simultaneously building its non-OP donor base. GiveWell treats this as an "exit grant" and is not factoring further Coefficient Giving funding beyond 2025 into plans; it had $527M available for granting at the start of 2024.⁴¹
• ARCOrganizationAlignment Research CenterComprehensive reference page on ARC (Alignment Research Center), covering its evolution from a dual theory/evals organization to ARC Theory (3 permanent researchers) plus the METR spin-out (Decembe...Quality: 57/100 proactively returned its $1.25M FTX Foundation grant and restructured: ARC Evals was spun out as METROrganizationMETRMETR conducts pre-deployment dangerous capability evaluations for frontier AI labs (OpenAI, Anthropic, Google DeepMind), testing autonomous replication, cybersecurity, CBRN, and manipulation capabi...Quality: 66/100, an independent nonprofit, in December 2023.¹⁸
• Effective Ventures planned to decentralize by offboarding 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, 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, and 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 as independent legal entities following its settlement.³ The EV decentralization plan has since been substantially executed: 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 completed its spin-out from EV in October 2025;¹⁴ EA Funds and CEA announced a combined spin-out with a target date of July 1, 2025;¹³ and 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's spin-out status as of this writing is not documented in publicly available sources.

Whether these measures have meaningfully reduced systemic concentration risk is debated within the EA community, given that Coefficient Giving remains the dominant funder for many organizations that have diversified away from FTX-linked sources. EA Forum analysis found that approximately half of Long-Term Future Fund funding continues to come from Coefficient Giving, illustrating that post-FTX diversification efforts have not eliminated funder concentration.²⁴

Donor Behavior

EA Funds experienced its first month-over-month donor decline in December 2022, after unbroken growth since 2017. Donor participation declined approximately 10–20% in 2023 relative to 2022. Giving What We Can pledge data showed approximately a 70% decline in new monthly pledges in 2023 versus 2022.²⁸

Perspectives on the FTX Lessons

Different observers have drawn varying conclusions from the FTX experience:

Concentration risk as central lesson: The dominant view within EA, reflected in post-FTX grantmaking guidelines and organizational finance discussions, holds that the collapse demonstrated the systemic risk of single-funder dependence and the need for reserve-building and source diversification.

Recovery as demonstration of resilience: Some observers note that the EA ecosystem has continued to function since 2022 despite the loss of a major funder, and that Coefficient Giving's gap-filling response demonstrated meaningful ecosystem resilience. From this view, the ecosystem's adaptation demonstrates capacity to absorb funding shocks without permanent structural damage.

Tension with hits-based giving logic: The recommendation to diversify funding sources is in tension with EA's own "hits-based giving" arguments, which hold that philanthropic resources should be concentrated in the highest-expected-value opportunities rather than spread for diversification's sake. Academic research on nonprofit funding finds that revenue concentration can benefit organizational efficiency and reduce administrative costs, while diversification can cause mission drift by creating conflicting demands from heterogeneous stakeholders (Frumkin and Keating 2011; Young 2008).⁴² Some analysts argue that applying diversification logic to funding sources (rather than to grant portfolios) is a distinct question that does not necessarily conflict with hits-based giving.

Reputational damage scope disputed: Some observers argue that reputational damage to EA from FTX has been overstated, noting continued organizational activity and new talent recruitment since 2022. The Hollywood Reporter noted in 2025 that EA had experienced a partial public image recovery, with continued EA organizational growth and new program launches.² Others argue the primary damage was to EA's credibility in policy and academic contexts. No systematic survey data or content analysis of this question has been published; the characterization of any "recovery" reflects qualitative assessments rather than measured outcomes.

Proposed Responses and Measures

The following summarizes responses that have been proposed or adopted by EA community members and organizations. These are descriptive of community discussions, not prescriptions.

Adopted by Organizations

Several specific measures have been publicly documented:

• Reserve targets: Multiple EA organizations (Rethink Priorities, MIRI) have adopted explicit minimum reserve policies of 12–24 months of operating expenses.³⁸³⁹
• Funder governance separation: EAIF separated Coefficient Giving employees from its leadership to reduce structural dependence.⁴⁰
• Multi-year funding commitments: GiveWell and Coefficient Giving structured a $300M commitment over three years to provide planning predictability.⁴¹
• Voluntary returns: ARC returned FTX funds before being legally compelled.¹⁸
• Legal independence: 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 completed its spin-out from EV in October 2025; CEA and EA Funds announced a combined spin-out targeting July 2025.¹³¹⁴

Proposed by Community Members

• Pre-liquidity diversification: EA Shareholder Diversification from AnthropicE851This article documents Jeffrey Epstein's financial relationships with AI researchers including Ben Goertzel, Marvin Minsky, Joscha Bach, Seth Lloyd, Nick Bostrom, and MIRI, drawing on DOJ files and...Quality: 60/100 documents proposals for diversifying at least a portion of EA holdings pre-IPO rather than waiting for full liquidity.
• Ecosystem reserve funds: Proposals for maintaining a reserve of 2–3 years' ecosystem funding in diversified assets, separate from active grantmaking pools.
• Committed-capital accounting standards: Proposals that EA organizations formally distinguish between "received" and "committed" funding in financial planning, treating only received funds as operational budget—directly addressing the failure mode that caused the most operational disruption in 2022. This principle is reflected in the organizational resilience measures adopted by Rethink Priorities, which explicitly excludes committed-but-unreceived funding from its operational plans.³⁸

Unresolved Questions

• Whether the diversification measures adopted since 2022 have materially reduced ecosystem concentration risk, given Coefficient Giving's continued dominant position
• Whether hedging instruments (e.g., options strategies, insurance products) applicable to EA organizations' concentrated exposure have been seriously evaluated
• Whether governance reforms to Future Fund-style grantmaking structures have been implemented by new EA-adjacent funders

The Anthropic Stake Recovery

One notable outcome of the FTX bankruptcy: FTX's $500M investment in Anthropic's Series B (April 2022, at a $4B valuation) became one of the more valuable assets in the bankruptcy estate.

The proceeds went entirely to FTX creditors, not to EA causes—illustrating that a successful investment fails to serve its intended philanthropic purpose when the holding entity collapses before capital is transferred.⁴ At Anthropic's current $380BValuation$380 billionAs of: Feb 2026Series G post-money valuation; second-largest venture deal ever behind OpenAI's $40BSource: reuters.comanthropic.valuation → valuation, the original stake (before subsequent-round dilution) would be substantially larger in notional value.

Key Uncertainties

External Effects

The FTX collapse affected EA's relationships with external institutions in ways that are partially documented, though no systematic study across institutions has been conducted.

Academic programs and university EA groups: Data from the Netherlands shows EA intro fellowship completions dropped from approximately 291 in 2022 to approximately 212 in 2023. A May 2023 survey of university group organizers conducted by CEA's groups team found that the FTX collapse significantly affected organizers' mental and emotional health, trust in EA leadership, and certainty in EA as a movement. University group organizers reported less interest in EA programs generally, though more interest in AI safety programs specifically.⁴³ More than $13M in Future Fund grants had gone to university-linked research projects; at least 20 researchers at Cornell, Princeton, Brown, and Cambridge received individual grants over $100K that were not honored.¹⁷

Outcomes for affected researchers: Media reporting documents that Future Fund grantees were left in difficult positions—some facing clawback concerns if they had received funds, others without funding they had planned around. At least one PhD candidate at Cambridge left their program when committed funding never arrived.¹⁷ Coefficient Giving opened applications to Future Fund grantees in November 2022, explicitly including those who had not received some or all of their committed funds.⁶ The National Science Foundation announced a $20M grant pool for AI safety research following FTX's collapse, with grants up to $800K.⁴⁴ Whether individual researchers at Cornell, Princeton, Brown, and Cambridge secured alternative funding, abandoned projects, or found other resolution has not been systematically documented in publicly available sources.

Policy credibility: CSET, the largest single recipient of EA-associated funding (over $100M from Coefficient Giving), publicly distanced itself from EA by November 2023, with its executive director stating "CSET is not an EA organization."²⁶ Whether other policy-facing organizations similarly distanced themselves has not been systematically documented; no published study of this effect across AI policy or biosecurity organizations exists.

Media framing: Coverage of EA shifted from predominantly positive to predominantly critical in major outlets during late 2022 and 2023, with articles in the New York Times, the Guardian, the Washington Post, New York Magazine, and the Economist characterizing EA in terms including "morally bankrupt," "ineffective altruism," and "defective altruism."²⁵ Some recovery in framing was noted by 2025, with The Hollywood Reporter describing EA's "Post-SBF Comeback."² No formal content analysis or computational sentiment study of this shift has been published.

Philanthropic sector: Anthropic CEO Dario AmodeiPersonDario AmodeiComprehensive biographical profile of Anthropic CEO Dario Amodei documenting his competitive safety development philosophy, 10-25% catastrophic risk estimate, 2026-2030 AGI timeline, and Constituti...Quality: 41/100 publicly distanced himself from the EA movement after the collapse, expressing concern that EA's institutional culture can be insular and that its emphasis on abstract long-term reasoning can come at the expense of practical judgment, per New York Times reporting.²⁷ No major mainstream philanthropic institutions have been identified in published reporting as formally ending engagement with EA as a direct consequence of FTX. The Stanford Social Innovation Review's critique of EA (characterizing it as "charitable imperialism") predates FTX and was separately motivated.⁴⁵

Limitations of This Analysis

The committed/disbursed distinction matters. The $160M figure represents commitments, not confirmed disbursements. The actual amount distributed before collapse was likely lower; Science magazine reported $132M in awards as of June 2022, though how much had been physically transferred to grantees is not established from public sources.⁸ This is the operationally critical figure for understanding actual losses to grantees, and it remains unresolved.

The 35% and 80–90% concentration figures have soft evidentiary basis. The estimate that FTX and Alameda co-founders' wealth made up approximately 35% of all EA funding comes from an EA Forum post without described methodology.²⁰ The 80–90% figure for combined Moskovitz/SBF dominance is not independently cited and should be treated as a rough community estimate.

Counterfactual analysis is not possible. We cannot establish what EA funding would look like without the FTX collapse. Some organizational restructuring (e.g., EV decentralization) may have had independent justifications and occurred partly because of the crisis.

Survivorship bias applies. This analysis focuses on organizations that survived. Projects and organizations that were abandoned may have had significant value that is difficult to quantify.

The FTX-Anthropic analogy has limits. Anthropic is a legitimate technology company, not a fraudulent exchange. The relevant parallel is concentration risk, not fraud risk. Drawing too strong an analogy between the two cases risks mischaracterizing the nature of the current risk.

Political dynamics are omitted. SBF's political donations and their aftermath are a significant part of the FTX story but are outside the scope of this funding-focused analysis. The Protect Our Future PAC donations appear in the funding table for completeness but are not analyzed here.

Attribution of organizational contractions is uncertain. The 2022–2023 period saw multiple EA funding sources contract simultaneously (Coefficient Giving's asset base also declined due to Moskovitz stock prices). Isolating FTX-specific effects on specific organizations' staffing from these broader funding retrenchments is not possible from public information.

Clawback resolution data is incomplete. The final resolution of clawback proceedings against EA organizations beyond EV—including the CFAR/Lightcone adversarial proceeding and the unnamed nonprofit's $5M case—is not publicly documented as of this writing. The statute of limitations expired November 2024, and FTX's full-creditor-repayment outcome may have rendered some proceedings moot, but no public confirmation of case dismissals or settlements for these specific proceedings has been found.

EV decentralization status is partially resolved. 80,000 Hours completed its spin-out in October 2025; CEA and EA Funds announced a combined spin-out targeting July 2025. The status of Giving What We Can's spin-out is not documented in publicly available sources as of this writing.