LLM watermarking
Linguistics-Aware Non-Distortionary LLM Watermarking
New research proposes a linguistics-aware watermarking method for LLMs that embeds detectable signals into generated text without distorting output quality, advancing AI content authentication.