Selective Disclosure Watermarking for Large Language Models
This addresses privacy concerns in multi-bit watermarking for LLMs by enabling fine-grained access control, which is a novel capability for the field.
The paper introduces Hierarchical Vocabulary Routing (HeRo), a watermarking framework for LLMs that enables selective disclosure of embedded metadata, allowing different verifiers to decode only authorized portions of the payload. Experiments show it maintains text quality, achieves high detection accuracy, and low latency.
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.