# An Ensemble Framework for Unbiased Language Model Watermarking

arXiv: https://arxiv.org/abs/2509.24043 · HTML: https://mlsanity.com/paper/2509.24043/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security.
- Builds on 20 of its 27 references that are in the map.
- First posted on arXiv 28 Sep 2025.

## Placed in

- LLM safety and security / [Watermarking generated content and models](https://mlsanity.com/safety/watermarking-generated-content-and-models/)

## Builds on (most cited first, 20 of 20)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://mlsanity.com/paper/1910.13461/) (2019)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [HuggingFace's Transformers: State-of-the-art Natural Language Processing](https://mlsanity.com/paper/1910.03771/) (2019)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Get To The Point: Summarization with Pointer-Generator Networks](https://mlsanity.com/paper/1704.04368/) (2017)
- [Teaching Machines to Read and Comprehend](https://mlsanity.com/paper/1506.03340/) (2015)
- [Multilingual Denoising Pre-training for Neural Machine Translation](https://mlsanity.com/paper/2001.08210/) (2020)
- [A Watermark for Large Language Models](https://mlsanity.com/paper/2301.10226/) (2023)
- [Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense](https://mlsanity.com/paper/2303.13408/) (2023)
- [Robust Distortion-free Watermarks for Language Models](https://mlsanity.com/paper/2307.15593/) (2023)
- [Undetectable Watermarks for Language Models](https://mlsanity.com/paper/2306.09194/) (2023)
- [Unbiased Watermark for Large Language Models](https://mlsanity.com/paper/2310.10669/) (2023)
- [Mark My Words: Analyzing and Evaluating Language Model Watermarks](https://mlsanity.com/paper/2312.00273/) (2023)
- [Adaptive Text Watermark for Large Language Models](https://mlsanity.com/paper/2401.13927/) (2024)
- [WaterBench: Towards Holistic Evaluation of Watermarks for Large Language Models](https://mlsanity.com/paper/2311.07138/) (2023)
- [Improved Unbiased Watermark for Large Language Models](https://mlsanity.com/paper/2502.11268/) (2025)
- [De-mark: Watermark Removal in Large Language Models](https://mlsanity.com/paper/2410.13808/) (2024)

## Built on by (most cited first, 6 of 6)

- [MC2Mark: Distortion-Free Multi-Bit Watermarking for Long Messages](https://mlsanity.com/paper/2602.14030/) (2026)
- [More Haste, Less Speed: Weaker Single-Layer Watermark Improves Distortion-Free Watermark Ensembles](https://mlsanity.com/paper/2602.11793/) (2026)
- [SharedRequest: Privacy-Preserving Model-Agnostic Inference for Large Language Models](https://mlsanity.com/paper/2606.05004/) (2026)
- [Tracing Provenance and Detecting Tampering with Complementary LLM Watermarks](https://mlsanity.com/paper/2608.12713/) (2026)
- [A Locally Tokenized Generative Model for Robust Time-Series Watermarking](https://mlsanity.com/paper/2608.19727/) (2026)
- [TTMark: Pairwise Distortion-Free Watermarking Beyond Single-Token Entropy](https://mlsanity.com/paper/2609.36372/) (2026)

## Abstract (arXiv)

As large language models become increasingly capable and widely deployed, verifying the provenance of machine-generated content is critical to ensuring trust, safety, and accountability. Watermarking techniques have emerged as a promising solution by embedding imperceptible statistical signals into the generation process. Among them, unbiased watermarking is particularly attractive due to its theoretical guarantee of preserving the language model's output distribution, thereby avoiding degradation in fluency or detectability through distributional shifts. However, existing unbiased watermarking schemes often suffer from weak detection power and limited robustness, especially under short text lengths or distributional perturbations. In this work, we propose ENS, a novel ensemble framework that enhances the detectability and robustness of logits-based unbiased watermarks while strictly preserving their unbiasedness. ENS sequentially composes multiple independent watermark instances, each governed by a distinct key, to amplify the watermark signal. We theoretically prove that the ensemble construction remains unbiased in expectation and demonstrate how it improves the signal-to-noise ratio for statistical detectors. Empirical evaluations on multiple LLM families show that ENS substantially reduces the number of tokens needed for reliable detection and increases resistance to smoothing and paraphrasing attacks without compromising generation quality.