# LLMs respond differently to harmful prompts when AI watermarking is used

> **Open Intelligence Dossier** · First detected: 2026-09-18 19:40 UTC · Category: Technology

---
*Canonical Source: https://pulse.byoviral.com/trend/2026-09-18/llms-respond-differently-to-harmful-prompts-when-ai-watermarking-is-used*