Holy crap, is Anthropic playing both sides? Totally exposed Turns out Dustin Moskovitz was an early investor in Anthropic back in 2021. Later, he donated his Anthropic equity to his charitable foundation, which Forbes estimated to be worth around $500 million in 2025. Based on the disclosed ownership cap, as Anthropic’s valuation skyrocketed, the theoretical value of this asset has now reached nearly $7.7 billion. And here’s where it gets interesting. Moskovitz’s organizations—Good Ventures, Coefficient Giving, and related donation channels—have been funding a bunch of AI safety institutions for a long time. These include safety research and evaluation groups like METR and Redwood, as well as projects like Tarbell, which trains AI-focused journalists. Tarbell’s graduates have gone on to work at outlets like TIME, The Verge, and the LA Times, consistently reporting on AI risks. Meanwhile, METR handles “third-party safety evaluations” for cutting-edge models from companies like Anthropic and OpenAI. So the whole chain looks like this: Anthropic becomes more valuable → Moskovitz’s donated Anthropic equity becomes more valuable → The AI safety community gets more funding → More research and media narratives emerge about AI being dangerous and needing stricter regulation → Governments increasingly rely on so-called independent third-party evaluations → Institutions like METR gain more power. And just recently, Dario Amodei started pushing for embedded third-party evaluators, even directly citing METR as an example. This means that in the future, these organizations might not just write reports—they could actually step into labs and decide which models are too dangerous or which companies should face restrictions. This is why Kevin Bass is now calling on the U.S. Congress to investigate. He’s not accusing Anthropic of secretly paying METR for favorable results—there’s no evidence of that yet. The real issue is this: the people funding the research, the companies being evaluated, the ones assessing the risks, and those spreading the “AI apocalypse” narrative are all part of the same interconnected network of money and relationships. There’s even personnel overlap. Members of Anthropic’s safety team can move to METR, Redwood gets funding from the same sources and subcontracts investigations to METR, and early Anthropic investors and observers are also tied to the funding side of these AI safety organizations. Basically, it’s like an early shareholder of a pharmaceutical company making a fortune, then using that money to fund drug safety agencies, medical media, and regulatory research—only for those same agencies to be recommended as “independent referees.” Even if no one is cheating, the structure itself is already awkward enough. Kevin Bass calls it a “regulatory capture machine”: First, tell everyone AI is dangerous. Then fund research proving AI is dangerous. Then train media to keep talking about how dangerous AI is. Finally, push governments to establish stricter regulations. And then let institutions within the same circle decide who gets to keep training AI.