Necessary and Sufficient Watermark for Large Language Models
Abstract: Large language models (LLMs) can now generate texts that are indistinguishable from those written by humans. Such remarkable performance of LLMs increases their risk of being used for malicious purposes. Thus, it is necessary to develop methods for distinguishing texts written by LLMs from those written by humans. Watermarking is one of the most powerful methods for achieving this. Although existing methods have successfully detected texts generated by LLMs, they inevitably degrade the text quality. In this study, we propose the Necessary and Sufficient Watermark (NS-Watermark) for inserting watermarks into generated texts with minimum text quality degradation. More specifically, we derive minimum constraints required to be imposed on the generated texts to distinguish whether LLMs or humans write the texts, and we formulate the NS-Watermark as a constrained optimization problem. Through the experiments, we demonstrate that the NS-Watermark can generate more natural texts than existing watermarking methods and distinguish more accurately between texts written by LLMs and those written by humans. Especially in machine translation tasks, the NS-Watermark can outperform the existing watermarking method by up to 30 BLEU scores.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Vlad_Niculae2
Submission Number: 3574
Loading