If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All

Eliezer Yudkowsky and Nate Soares's If Anyone Builds It, Everyone Dies is a New York Times bestseller that presents the case for existential risk from artificial superintelligence in accessible, urgent prose. Drawing on decades of work at the Machine Intelligence Research Institute, the book distills the core arguments of AI alignment into a form that has reached a broad public audience. The volume captures the rationalist and effective altruist worldview that has shaped AI safety discourse since approximately 2015, and key executives at major AI laboratories — including Anthropic's Amodei, OpenAI's Sutskever, and DeepMind's Hassabis — have all engaged substantively with Yudkowsky's arguments.

Brought out in September 2025, the book arrives at a moment when artificial intelligence has moved from academic speculation to the center of global policy debate. Yudkowsky effectively founded the field of AI alignment, and his earlier writings — the 'Sequences' published on LessWrong between 2006 and 2009 — helped shape the rationalist movement that would go on to influence Silicon Valley's approach to technology and risk. Effective Altruism and rationalism became dominant ideological frameworks within AI laboratories, most notably at Anthropic, founded by former OpenAI researchers citing safety concerns. The 2023 open letter on AI extinction risk, signed by hundreds of researchers and industry leaders, reflected decades of advocacy by Yudkowsky and his intellectual community. Whether he proves to be a prescient prophet or an overcautious catastrophist, his influence on how the AI industry conceives of its own responsibilities is beyond dispute.