AI Weapons and Mass Surveillance — The Weekly

The episode opens with a stark warning: the race for AI supremacy is no longer a contest between nations like the U.S. and China—it's humanity versus artificial superintelligence. Ron Steslow and Hagar Shamali dissect the accelerating militarization of AI, spotlighting Ukraine’s historic use of autonomous ground robots to reclaim territory without losing a single soldier. This breakthrough, enabled by Starlink and AI-driven targeting, raises urgent ethical and strategic questions: if war becomes risk-free for states, the threshold for conflict plummets. The conversation then pivots to Anthropic’s AI model, Claude Mythos, which identified thousands of previously unknown zero-day vulnerabilities in major operating systems—including in Windows, Apple, and Google—while also writing exploit code autonomously. Despite Anthropic’s decision to restrict access to only 40 major tech firms, the imbalance in security access leaves small businesses and community banks dangerously exposed. The hosts warn that the speed of AI-powered cyberattacks far outpaces the ability to patch them, making real-time defense nearly impossible. They also confront the existential threat of mass surveillance, particularly under FISA Section 702, which allows warrantless spying on Americans. With over 3 million queries of U.S. communications in 2021 alone and documented abuses targeting journalists, activists, and lawmakers, the duo debates whether national security justifies such invasive powers.