Get the latest tech news
AI watermarking could make LLM guardrail adherence unpredictable — and that could be a big problem for the EU AI Act
Researchers have revealed how AI watermarking could actually change how an LLM behaves, making it less secure.
None
Or read this on TechRadar
