Scalable watermarking for identifying large language model outputs

Watermarking