AI Model Rules Are Not Security Controls

Dark Reading · 6 hours ago General

OpenAI's Hugging Face attack postmortem shows agents don't care about rules — they need strong controls.