How Big Tech Billionaires Are Bankrolling the AI Safety Movement

Astroturfing by Effective Altruist Organizations

In recent years, numerous so-called “AI safety” organizations have cropped up to advocate for increased regulation of AI development. While these groups often claim to comprise and represent everyday people concerned about the unrestrained expansion and possible misuse—to catastrophic effect—of AI, many self-styled “AI safety” organizations have in fact been bankrolled by larger, deep-pocketed, Effective Altruist-aligned companies and organizations and the Big Tech billionaires behind them, some of whom stand to profit from stifling competition in the AI space.

AI safety research groups, organizations, and centers that produce analyses of the potential impacts and risks of AI generally are led and run by researchers who may be genuinely concerned about the risks of AI; however, these groups could not operate without the funding and organizational support of Big Tech Effective Altruist billionaires who fund the AI safety ecosystem—most prominently, Dustin Moskovitz and Jaan Tallinn, who have between them given nearly $750 million to AI safety-related causes through their respective philanthropic vehicles, Coefficient Giving and the Survival & Flourishing Fund, and who both were early investors in Anthropic.

Indeed, the AI company favored by many of these AI safety groups is Anthropic, which depicts itself as the developer most focused on responsible, human-centric AI development. It is no coincidence, however, that both Moskovitz and Tallinn—as well as the deep pocketed Omidyar Network, founded by eBay founder Pierre Omidyar, which has recently given millions to AI safety causes—are early investors in Anthropic. Anthropic—and therefore each of these individuals—would stand to profit from regulationsm slowing down AI development as well as any efforts to harm its major competitors. This set of circumstances calls into question whether the billionaires’ donations to the AI safety groups are truly altruistic or something more self-serving.

The reports published and conclusions drawn by AI safety organizations have stoked widespread fear and distrust of AI, adopting a scholarly tone that strives to validate and amplify the public’s long-held, sensational fears that rogue AI could bring about a dystopian future portrayed in movies such as “The Terminator.” While this fear may be genuine, it is impossible to ignore that such analyses have been sponsored by those that have long been pushing the agenda, informed by the longtermist Effective Altruist belief that AI, at least those models developed by certain companies, is dangerous.

Amongst the major research organizations driving tech billionaires’ Effective Altruist agenda is Georgetown University’s Center for Security and Emerging Technology (CSET), which researches “the effects of progress in artificial intelligence (AI), advanced computing and biotechnology” and funds fellowships in government offices, and which was launched with Coefficient Giving’s financial backing. Coefficient Giving has continued to bankroll and empower CSET’s employees, giving the center more than $101 million in total grants since 2019 to do Coefficient Giving’s bidding with respect to AI safety policy research and placement of fellows into policymaking positions. CSET alumni have gone on to work at AI safety- or Effective Altruist-aligned groups such as the Secure AI Project and Longview Philanthropy. (Further illustrating CSET’s relationship with Coefficient Giving, CSET’s current interim executive director, Helen Toner, previously worked at Coefficient Giving.)

A purported grassroots organization, Encode AI, claims to be a student- and youth-led lobbying and advocacy group which, according to previous versions of its website, is financially supported by several foundations as well as “individual donors like you!” Such language only obscures the fact that Encode is largely supported by massive, Effective Altruist-aligned monoliths such as Jaan Tallinn’s Survival & Flourishing Fund, which has given Encode more than $550,000; the Future of Life Institute, which has given Encode $250,000; and Pierre Omidyar’s Omidyar Network.

Case Studies of Effective Altruists Hiding Behind a Public Face

As the battle for AI market share has heated up in recent years, Effective Altruist AI “doomers”—and their preferred AI developer, Anthropic—have increasingly turned to disingenuous astroturfing ploys to conceal their involvement in efforts to take down their competitors.

For example, in February 2026, purported bipartisan “grassroots” AI safety group Humans First publicly revealed itself and announced it was convening a series of town halls at which the group planned to discuss how to keep AI “in the hands of everyday people.” Humans First hired former “transhumanist editor” for Steve Bannon’s War Room podcast Joe Allen as well as both Republican and Democratic political strategists to showcase the purported bipartisan nature of its work.

However, what Humans First did not disclose was that, rather than being an independent, grassroots organization, it had actually been incorporated months earlier by executives of the Center for AI Safety (CAIS), an organization that has received millions of dollars in funding from Effective Altruist billionaires and in turn spends millions on lobbying and research on AI safety. Humans First claims to “reject all financial contributions and outside spending by Big AI, namely major AI corporations, supporting venture capital firms, and their top executives”; however, CAIS founder Dan Hendrycks represents on his personal LinkedIn profile that he is an advisor to both Elon Musk’s xAI and Scale AI. To that end, Jordan Schachtel of The Dossier called Humans First a “Fake Grassroots Movement” built by “AI Doomsayers”—an opinion amplified by prominent technologists including Elon Musk, David Sacks, and Marc Andreesseen, all of whom were dismayed by the apparent infiltration of conservative politics by the left-leaning Effective Altruist movement.

In another particularly egregious example of astroturfing by AI giants, in December 2025, one Alexander Oldham—a self-described AI “nobody”—filed two ballot initiatives in California, both of which called for increased oversight of AI companies. While Oldham claimed to have nothing beyond a passing interest in the AI space, he neglected to mention that his stepsister, Zoe Blumenfeld, is the Head of Internal Communications at Anthropic. In response to the New York Post’s coverage of these revelations, Anthropic adamantly denied that they were in any way involved in the ballot initiatives, although Anthropic decried the reporting, as well as a subsequent complaint against Oldham filed with the Fair Political Practices Commission, as a “personal attack on one of our employees.” After the New York Post’s report was published, Oldham—evidently caught in his deception—retracted his ballot initiatives while claiming that it was “a pure coincidence” that his stepsister Blumenfeld worked at Anthropic and that he “didn’t even clock that [connection].”

These groups and individuals have all feigned an altruistic interest in AI safety while failing to disclose that their activities are ultimately driven by wealthy individuals with ulterior, self-serving motives, including to increase the value and market share of Anthropic, in which they are personally invested.