Your daughter’s smile is dearer to you than that of a stranger half a world away. Of course it is. Human life is situated and embodied, and so are our obligations and loves. But the ideology that holds sway over many Silicon Valley elites diagnoses this preference for those near and dear to you as an evolutionary defect. Instead, according to the adherents of this worldview, every sentient being in every place, in every time, holds the same moral claim on every other. This philosophy is known as Effective Altruism (EA).
Scandals in the EA world, most notably the downfall of the cryptocurrency entrepreneur Sam Bankman-Fried, prompted many effective altruists to renounce the name in recent years. But the ideas endure. In 2025, six EA-aligned grantmakers deployed $1.25 billion, and this number is set to grow along with the success of the world’s leading artificial intelligence company, Anthropic, which was founded on EA principles. The firm’s founders have pledged to donate 80 percent of their wealth upon the firm’s IPO. These wealth pledges, combined with the company policy of matching employee donations, will create billions in philanthropic capital, channeled through EA’s entrenched institutions.
Effective Altruism is the final flowering of utilitarianism, that Victorian philosophy of calculating reformers, transplanted into Silicon Valley boardrooms and granted the false vitality of billion-dollar balance sheets. Because it is constitutively blind to what makes a civilization worth preserving, its utopia will entail dissolving America in the name of humanity.
“Several historical developments help explain EA’s success.”
Several historical developments help explain EA’s success. One is the rise of mass media, which widened the circle of concern that once might have encompassed only family and neighbors. By the beginning of the 1970s, televisions were present in nearly every American home. Footage of people suffering abroad became a regular fact of American life. In 1972, the philosopher Peter Singer published “Famine, Affluence, and Morality,” the essay that would become EA’s foundation. His argument was simple: If we can prevent suffering at little cost to ourselves, we must, and it makes no moral difference whether the person we help is “a neighbor's child ten yards from me” or a stranger on another continent.
Another precondition for EA’s rise was the modern pathologization of traditional sources of identity––nation, religion, family, tradition. Especially after World War II, such particular loyalties were recast as the wellspring of fascism. Later generations, whose education dispossessed them of everything particular, were prepared to receive the EA program.
In his 1981 book After Virtue, Alasdair MacIntyre argued that every culture is defined by its stock of “characters,” the social roles that embody its moral ideals. “Characters,” he wrote, “are the masks worn by moral philosophies.” The twentieth century’s main character was the bureaucrat, whose authority was derived from a seemingly scientific knowledge of human affairs, grounded in the supposed incommensurability of facts and values. The bureaucrat administers according to facts, while values are condemned to the private sphere. His expertise cannot be challenged. The standard of his authority is effectiveness.
He flattens all goods into a single quantifiable dimension, converting them into units like lives saved, suffering averted, and dollars per outcome. The units’ apparent neutrality conceals a range of ideological assumptions. The human “good” becomes a function of EA’s calculus. Then, pretending value- neutrality, EA denounces defection as backwards, unscientific, and immoral. Any challenge appears to spurn both science and human dignity.
More recently, EA’s impartialism has expanded far beyond those currently living, to what “longtermists” envision as 10^58+ potential future beings––human and digital. These future beings’ existence depends on the avoidance of human extinction. The longtermists consider AI the dominant existential risk, and frontier development is concentrated in a few American labs.
Some have taken the argument to its violent conclusion. Eliezer Yudkowsky, the most influential of the doomsayers, wrote in Time that nations should “be willing to destroy a rogue datacenter by airstrike,” nuclear war notwithstanding. And since 2022, the Zizians, a cult that splintered off from the same Bay Area rationalist scene, have been linked to six killings.
Congress must treat the philanthropic apparatus of frontier AI as it treats any concentration of unaccountable power. Lawmakers should demand disclosure of the grantmaking networks that fund labs with national security implications. They should condition federal contracts on terms any defense contractor would recognize: Disclose the instructions and value systems trained into the models, certify systems sold to the government will execute lawful orders without an engineer's veto, and reserve the government’s right to audit.
The machines our government buys must serve the American people, not some present or future humanity. And procurement standards should favor open-weight models, which anyone can download, run, and test, so that the nation never depends on intelligence a handful of labs can withhold.