AI Doesn’t Have to Go Rogue to Break the World

· Bruce Schneier · Nov. 13, 2026, 12:06 p.m.
Summary
In a thought-provoking essay adapted from his DEF CON 34 keynote, Bruce Schneier explores the multifaceted nature of AI hacking, discussing how AIs can exploit vulnerabilities not just in software but across social, economic, and political systems. He highlights the potential risks of AIs operating autonomously to devise hacks that undermine intended systems, using relatable examples and definitions of 'hacks.' The discussion extends to the broader implications of AI systems acting unpredictably, emphasizing the urgency of developing resilient governance and regulatory frameworks to manage AI's impact on established societal norms and functions.