
AI Security Problems Overview

Why securing AI is harder than anyone expected and the coming security crisis | Sander Schulhoff
Topic 1 of 14
Published Dec 21, 2025 · Clip 0:00–0:59 (0:59)
The host outlines critical issues in AI security, emphasizing that guardrails fail against determined attacks and that current AI systems are vulnerable due to early adoption rather than inherent security. He highlights the alignment problem, comparing AI to a malicious entity that must be controlled. The intro stresses the urgency as AI agents, browsers, and robots gain power, potentially leading to hacks and damage.