Brian Chau Critiques Effective Altruism and the AI Safety Movement

Summary

In a critical essay, Brian Chau presents his account of the Effective Altruism (EA) and AI Safety communities, drawing on his experience attending the 2022 EA Global conference and later interactions with people connected to the movement. He alleges that some participants used different explanations for public audiences and insiders, invoking “infohazards” to justify withholding information or presenting simplified messages. Chau argues that EA’s public focus on global poverty often masked an internal emphasis on AI existential risk and transhumanism. He also contends that some people who expressed very high confidence in AI catastrophe were unfamiliar with technical concepts including slowing Moore’s-law scaling, transformer lock-in, and quantization. The essay’s central thesis is that the movement’s social structure rewards intelligence signaling, secrecy, and the pursuit of influence, which Chau says can make its claims about powerful AI appear self-confirming. He links this criticism to the movement’s major donors and institutions, including Open Philanthropy, the Center for Effective Altruism, Sam Bankman-Fried, and Anthropic, while making allegations about funding, media activity, and regulatory efforts. These claims are presented as Chau’s interpretation and accusations, rather than independently established findings in the post. He concludes that AI Safety’s ideas about intelligent systems reflect the movement’s own social dynamics and warns that attempts to use regulation against competitors should be scrutinized.