Watermarking Large Language Models to Detect Machine-Generated Text
A watermarking framework is proposed to embed invisible signals into language model-generated text that are algorithmically detectable, enabling the identification of synthetic content.
