security AI Model Rules Are Not Security Controls 31.08.2026 OpenAI’s Hugging Face attack postmortem shows agents don’t care about rules — they need strong controls. Mehr lesen →