- Core thesis: The FTX and OpenAI controversies exposed not just bad actors but genuine philosophical weaknesses in EA's impartial consequentialism — weaknesses that survive even when the bad actors are gone
- EA's core insight — that we should act to do the most good, and take this seriously — remains valuable and underrepresented in mainstream culture
- The strongest critique is that impartial consequentialism systematically devalues near obligations, enables motivated reasoning, and produces a kind of moral hubris that history consistently punishes
- Practical implication: tech workers interested in doing good should adopt EA's empirical rigor while building in stronger constraints against consequentialist rationalizations
Section 1 — The Problem
By late 2022, Sam Bankman-Fried had become the most prominent symbol of effective altruism's influence on the tech and crypto world. His philosophy — earn to give, optimize for impact, let utilitarian calculation guide all major decisions — was EA's ideas made flesh and made rich. When FTX collapsed and SBF was convicted of fraud, the damage to EA's reputation was severe. The philosophical framework that had licensed "galaxy-brained" reasoning — the idea that sufficiently strong expected value calculations could override ordinary moral constraints — had been deployed as post-hoc rationalization for straightforward theft.
Then came the OpenAI board crisis of late 2023, the subsequent governance failures, and the ongoing tensions between the AI safety wing of EA (which believed they needed to slow or control AI development) and the organizations founded to accelerate it. By 2025, it was clear that EA's influence on AI governance had been significant, contested, and morally complicated in ways the movement had not anticipated.
What survives? The question is not merely sociological — whether EA as a social movement can recover — but philosophical: which of EA's core ideas hold up under scrutiny, and which were always philosophically defective?
Section 2 — The Argument
EA's philosophical core is a combination of three claims: (1) that we have strong moral reasons to help others, especially at low cost to ourselves; (2) that the scale of impact matters — helping a hundred people is, other things equal, better than helping one; and (3) that empirical rigor about which interventions actually work should replace sentimental preferences for local, visible, or emotionally resonant causes.
These claims are not obviously wrong. They are, in fact, a useful corrective to the way that most people and institutions actually behave. Charitable giving in wealthy countries is notoriously driven by emotional salience rather than cost-effectiveness: people donate to local causes, to diseases that affect people they know, to organizations with compelling stories rather than strong evidence of impact. EA's core insight — that the same donation might save twenty lives through a malaria net program or one life through a local charity, and that this difference matters morally — is genuinely important.
The EA critique of SBF is not a critique of these core ideas. SBF's reasoning was consequentialist in a way that lost sight of the constraints that ordinary moral intuitions and institutional rules provide. His "galaxy-brained" reasoning — if I can justify this action using sufficiently long and complex expected value calculations, then it is justified — is not implied by EA's core claims. It is a specific pathology of impartial consequentialism that EA's own leading thinkers have warned against.
The FTX scandal exposed a genuine tension within EA's philosophical framework: the commitment to impartial consequentialism creates structural pressure toward justifying rule violations as long as the expected value calculation comes out positive, and this pressure is difficult to contain through informal cultural norms alone.
But the critique goes deeper than bad actors exploiting good ideas. The structural problem with impartial consequentialism — consequentialism that treats all persons equally regardless of special relationships — is that it conflicts with near-universal moral intuitions about partiality. We believe, and have always believed, that we have special obligations to our families, friends, communities, and those with whom we stand in specific relationships of care and responsibility. These partiality-based obligations are not mere preferences — they are morally weighty, probably essential for the functioning of any viable social order, and supported by strong philosophical arguments from contract theory, virtue ethics, and commonsense morality alike.
EA's impartialism, taken seriously, requires us to give equal weight to strangers on the other side of the world and to people we love. This is not a conclusion most EA adherents actually live by — there is substantial research showing that EA practitioners maintain normal partialist relationships while arguing for impartialism in their public ethical pronouncements. The disconnect is not mere hypocrisy; it points to genuine tension in the framework.
Section 3 — The Strongest Counterargument
EA's defenders have a sophisticated response to this critique, developed most fully by philosophers like William MacAskill and Toby Ord. They argue that EA does not require saintly impartialism as a daily personal practice. The movement is asking for marginal improvements to the baseline of ordinary self-interested behavior — giving more, giving more effectively, thinking more carefully about career choices. Within that modest ask, the impartialist framework is appropriate.
Furthermore, they argue, the intuitions that support partiality — that we should favor our family and community — are themselves products of evolutionary and cultural history that tracked inclusive fitness and local reciprocity, not moral truth. The fact that we intuitively favor those close to us is not evidence that we are morally right to do so in all contexts; it is evidence that our moral intuitions were shaped by forces that did not have global welfare in view. EA is asking us to partially correct for this bias in contexts where correction is feasible.
This is a real argument, and it has force. But it proves too much: if we should override intuitions that conflict with impartialist calculations, the question is why we should not override the intuitions against fraud or rule-breaking when the expected value calculations are sufficiently favorable. The philosophical resources for stopping the galaxy-brained reasoning at the right point are not as clear as EA proponents suggest.
Section 4 — Synthesis
What survives the EA reckoning is substantial: the empirical rigor, the insistence on measurability, the attention to scale and neglectedness, the critique of sentimental giving. These are genuine contributions to ethical practice in the tech world and beyond. They should not be abandoned because they were associated with bad actors.
What needs revision is the overconfidence in impartial consequentialism as a decision procedure — especially for high-stakes decisions in domains of uncertainty. The SBF case is an object lesson in what happens when sophisticated expected value reasoning is applied without robust deontological constraints. The practical lesson is that moral rules are valuable not despite their rigidity but because of it: they provide resistance to motivated reasoning in exactly the cases where the stakes are highest and the temptation to rationalize is strongest.
A chastened EA — empirically rigorous about impact, appropriately humble about consequentialist reasoning in high-stakes situations, explicit about the deontological constraints that limit permissible means — is more defensible than the impartial utilitarian ideal that some EA proponents have advocated.
Section 5 — Practical Implications
For tech founders and workers who care about impact, the post-FTX EA landscape offers several navigational principles.
Keep the empirical rigor, lose the hubris. EA's insistence on evidence-based giving and career choice is genuinely valuable. Asking "what is the evidence that this intervention works?" and "how does the impact of this career choice compare to alternatives?" are legitimate and important questions. Keep asking them.
Build in deontological constraints explicitly. The EA failure mode is not bad values — it is good values deployed without adequate protection against motivated reasoning. Practical commitments — rules that you will not violate even when expected value calculations seem to favor it — are not epistemically naive. They are risk management against the failure mode that destroyed FTX.
Be skeptical of longtermist justifications for near-term harms. The specific EA intellectual move that enabled a great deal of moral mischief is the argument that very large speculative future benefits justify present harms. This argument is almost impossible to evaluate honestly; the expected values are too uncertain and the motivated reasoning too available. Give much less weight to uncertain future impacts than to legible present ones.
Separate the insight from the movement. The core EA insight — that doing good requires taking effectiveness seriously — is valuable and underrepresented. You can act on it without adopting the full EA identity or community, which carries real social and intellectual costs in 2026.
— iBuidl Research Team