
Escalating Risks with Agents and Robots

Why securing AI is harder than anyone expected and the coming security crisis | Sander Schulhoff
Topic 7 of 14
Published Dec 21, 2025 · Clip 18:28–19:45 (1:18)
Sander agrees on rising risks with agents leaking data, costing money, or enabling harms like bioweapons. He warns of prompt-injected robots causing physical damage, citing jailbroken LLM-powered robots. The discussion stresses urgency as adoption grows beyond limited-damage chatbots.