Multi-Designated Detector Watermarking for Language Models

TL;DR

This paper introduces multi-designated detector watermarking for large language models, enabling secure, high-quality watermarked outputs detectable by specific detectors, with optional claimability for ownership assertion.

Contribution

It formalizes the concept of multi-designated detector watermarking, proposes a framework using multi-designated verifier signatures, and introduces claimability as a new security feature.

Findings

Effective watermark detection by designated detectors

No perceptible quality degradation in watermarked outputs

Flexible implementation with satisfactory performance

Abstract

In this paper, we initiate the study of \emph{multi-designated detector watermarking (MDDW)} for large language models (LLMs). This technique allows model providers to generate watermarked outputs from LLMs with two key properties: (i) only specific, possibly multiple, designated detectors can identify the watermarks, and (ii) there is no perceptible degradation in the output quality for ordinary users. We formalize the security definitions for MDDW and present a framework for constructing MDDW for any LLM using multi-designated verifier signatures (MDVS). Recognizing the significant economic value of LLM outputs, we introduce claimability as an optional security feature for MDDW, enabling model providers to assert ownership of LLM outputs within designated-detector settings. To support claimable MDDW, we propose a generic transformation converting any MDVS to a claimable MDVS. Our implementation of the MDDW scheme highlights its advanced functionalities and flexibility over existing methods, with satisfactory performance metrics.