Improved Unbiased Watermark for Large Language Models

TL;DR

This paper introduces MCmark, a novel unbiased watermarking method for large language models that improves detectability and robustness without affecting text quality, aiding in authenticating AI-generated content.

Contribution

We propose MCmark, a multi-channel unbiased watermarking technique that significantly enhances detection and robustness over existing methods in large language models.

Findings

Over 10% improvement in detectability compared to state-of-the-art methods

Preserves original language model distribution

Demonstrates robustness across various models

Abstract

As artificial intelligence surpasses human capabilities in text generation, the necessity to authenticate the origins of AI-generated content has become paramount. Unbiased watermarks offer a powerful solution by embedding statistical signals into language model-generated text without distorting the quality. In this paper, we introduce MCmark, a family of unbiased, Multi-Channel-based watermarks. MCmark works by partitioning the model's vocabulary into segments and promoting token probabilities within a selected segment based on a watermark key. We demonstrate that MCmark not only preserves the original distribution of the language model but also offers significant improvements in detectability and robustness over existing unbiased watermarks. Our experiments with widely-used language models demonstrate an improvement in detectability of over 10% using MCmark, compared to existing state-of-the-art unbiased watermarks. This advancement underscores MCmark's potential in enhancing the practical application of watermarking in AI-generated texts.