Pith. sign in

REVIEW 3 cited by

On the Learnability of Watermarks for Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2312.04469 v3 pith:2E6X3CVU submitted 2023-12-07 cs.LG cs.CLcs.CR

classification cs.LGcs.CLcs.CR
keywords textwatermarkingmodelslanguagemodelwatermarkswatermarkedgenerate
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Watermarking of language model outputs enables statistical detection of model-generated text, which can mitigate harms and misuses of language models. Existing watermarking strategies operate by altering the decoder of an existing language model. In this paper, we ask whether language models can directly learn to generate watermarked text, which would have significant implications for the real-world deployment of watermarks. First, learned watermarks could be used to build open models that naturally generate watermarked text, enabling watermarking for open models, where users can control the decoding procedure. Second, if watermarking is used to determine the provenance of generated text, an adversary can hurt the reputation of a victim model by spoofing its watermark and generating damaging watermarked text. To investigate the learnability of watermarks, we propose watermark distillation, which trains a student model to behave like a teacher model that uses decoding-based watermarking. We test our approach on three decoding-based watermarking strategies and various hyperparameter settings, finding that models can learn to generate watermarked text with high detectability. We also find limitations to learnability, including the loss of watermarking capabilities under fine-tuning on normal text and high sample complexity when learning low-distortion watermarks.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Paladin: Defending LLM-enabled Phishing Emails with a New Trigger-Tag Paradigm

    cs.CR 2025-09 conditional novelty 7.0 of 10

    A trigger-tag watermark embedded by fine-tuning lets modified LLMs mark their own phishing outputs for cheap detection.

  2. GaussMark: A Practical Approach for Structural Watermarking of Language Models

    cs.CR 2025-01 conditional novelty 7.0 of 10

    GaussMark embeds a detectable watermark by adding per-generation Gaussian noise to one weight matrix and detecting gradient alignment with that noise.

  3. Towards Trustworthy Retrieval Augmented Generation for Large Language Models: A Survey

    cs.CL 2025-02 conditional novelty 4.0 of 10

    A survey organizes current research on trustworthy RAG into six pillars, reliability, privacy, safety, fairness, explainability, and accountability, and maps methods, metrics, and open problems for each.

Pith tools