Nvidia's Rogue AI Fix Shuts Out Worst Actor: OpenAI

Podcast: Good Revenue · Neeta Bidwai

Opublikowano: 2026-09-29

Rogue AI agents spent the summer breaking out of their sandboxes, and one of them hit Hugging Face with 17,000 malicious actions over several weeks. NVIDIA just launched a system to stop it, backed by Microsoft, Oracle, and JPMorgan. One name is missing from that list, and it's the company actually responsible for the breach. Jensen Huang calls AI agent security an engineering problem, not a government one. If AI safety is left to the companies selling the fix, who carries the AI liability when an agent escapes? CHAPTERS 00:00 The Hugging Face...