EA Safety
Contraptions
Opinion — commentary, not a factual news event.
EA started as giving and career advice; now it’s a major force in AI safety. That matters because the movement is starting to need safety checks of its own.
Based on reporting by Contraptions — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
This essay argues that effective altruism has grown into something bigger and stranger than a charity philosophy. The author calls it the “EA formation”: a loose but powerful mix of philanthropy, rationalism, longtermism, existential-risk thinking, AI safety, AI governance, and parts of the frontier AI world. The point is not who carries the label. It’s who has been shaped by the same ideas and methods.
The core worry is simple enough to state and hard to dismiss. The movement is very good at assigning numbers to uncertainty, but much weaker at questioning the assumptions underneath those numbers. It can say 60 percent instead of 80 percent and still leave untouched the deeper questions: what counts as value, which futures matter, which evidence counts, and who gets to decide priorities.
That matters because this is not just a philosophy seminar. The formation exists to move money, careers, institutions and capital. And over time, AI has given it an almost theological charge: extinction, transcendence, a future so large that small probabilities can justify huge commitments, and a small group of people who may shape what comes next. In that setting, career choice becomes vocation, cause prioritization becomes moral duty, and AI alignment starts to look a lot like salvation.
The essay traces three ancestral strands. One is evidence-driven philanthropy, associated with GiveWell. Another is the consequentialist line running through Peter Singer, Toby Ord, William MacAskill, Giving What We Can and 80,000 Hours. The third is the rationalist, transhumanist and AI-risk cluster around LessWrong, Yudkowsky, MIRI, Bostrom and FHI. Those strands were already being brought together at the first EA Summit in 2013.
What changed was scale. Once future people are added to the moral ledger, the numbers get enormous. Once existential risk enters, protecting that future becomes urgent. Once transformative AI is treated as a possible source of that risk, AI can swallow the rest of the agenda. Global health and animal welfare still matter inside EA, but the essay’s argument is that the gravitational center has moved. The formation is no longer just trying to solve AI safety. It may now need EA safety too.
My take — AI-written commentary, not fact-checked reporting
This is the part where the nice people with spreadsheets start to look like a church with better fonts. The problem isn’t that they care about AI; it’s that any movement convinced it alone has the clean moral math eventually starts confusing confidence with wisdom. That’s how “reasonable” turns into a little too much power, very quickly.
Read more about this at: Contraptions