Book a meeting
Ars TechnicaArtificial Intelligence

LLMs respond differently to harmful prompts when AI watermarking is used