Bruce Schneier — Snowden, Crypto Wars, and the Future of Agentic Hacking
Summary
Security expert Bruce Schneier joins Hacked to discuss the dangers of autonomous AI agents, comparing their unpredictable behavior to the myth of King Midas. He warns that AI could exploit loopholes in non-computer systems like tax codes and argues that liability for AI actions should rest with owners, akin to dog bites. The conversation also covers the unsustainability of current AI business models, the rise of open-source models, and the ongoing "Crypto Wars" over government access to encryption.
Topics discussed
Intro: AI as a power enhancer and Bruce Schneier's background
EFF mission, leadership changes, and upcoming Black Hat talks
Agentic hacking: Genie-like behavior and the Hugging Face incident
VW diesel cheat analogy and AI finding tax loopholes
AI business models, open source vs. frontier, and China's strategy
Regulatory capture, liability, and human-in-the-loop decisions
Zuckerberg's regulatory strategy and AI copyright ownership
Sponsors: NordLayer and Notion custom agents
AI in mathematics: Solving problems and checking proofs
Blowfish encryption history and the Crypto Wars
Global push to weaken encryption and security theater
Snowden revelations, NSA supply chain attacks, and human targets
IoT security, European regulations, and packaging differences
Crypto skepticism: Blockchain inefficiency and stablecoin scams
Corporate personhood and 'Rewiring Democracy' book overview
Trust in the age of AI and closing remarks
Listen ad-free on Castria