Selective Disclosure Watermarking for Large Language Models
2026-07-06 • Cryptography and Security
Cryptography and SecurityArtificial IntelligenceComputation and LanguageMachine Learning
AI summaryⓘ
The authors present a new way to hide secret messages inside AI-generated text that lets different people see only parts of the hidden message based on their permission. Their method, called Hierarchical Vocabulary Routing (HeRo), splits the vocabulary into layers that carry different parts of the watermark. This approach keeps the text sounding natural and doesn’t reduce quality. Their tests show that HeRo can accurately detect hidden messages quickly while allowing for fine control over who sees what.
watermarkinglarge language modelsembedded metadataselective disclosurevocabulary partitioningsampling processaccess controldetection accuracytext quality
Authors
Xuyang Chen, Xiang Li, Yangxinyu Xie, Qi Long
Abstract
Watermarking methods embed imperceptible and verifiable signals into text generated by large language models (LLMs). Existing approaches include zero-bit schemes for distinguishing synthetic text from human writing and multi-bit schemes for embedding metadata. However, current multi-bit watermarking methods do not allow selective disclosure: verifying any part of the watermark requires revealing the entire embedded message. This lack of control leads to unnecessary information exposure and raises privacy concerns. We propose Hierarchical Vocabulary Routing (HeRo), a watermarking framework that enables selective disclosure of embedded metadata. The method recursively partitions the vocabulary and distributes watermark information across hierarchical layers, so that different verifiers can decode only the portions of the payload corresponding to their access level. We show that the proposed scheme preserves the unbiasedness of the underlying sampling process and thus maintains text quality. Experiments demonstrate that our framework supports fine-grained access control while achieving high detection accuracy and low latency. Code is available at https://github.com/xuyangc03/hero-watermark.