# Segmenting Watermarked Texts From Language Models

arXiv: https://arxiv.org/abs/2410.20670 · HTML: https://mlsanity.com/paper/2410.20670/

## Key facts

- Cited by 4 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 19 of its 37 references that are in the map.
- First posted on arXiv 28 Oct 2024.

## 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, 19 of 19)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [A Watermark for Large Language Models](https://mlsanity.com/paper/2301.10226/) (2023)
- [Robust Distortion-free Watermarks for Language Models](https://mlsanity.com/paper/2307.15593/) (2023)
- [On the Reliability of Watermarks for Large Language Models](https://mlsanity.com/paper/2306.04634/) (2023)
- [Adversarial Watermarking Transformer: Towards Tracing Text Provenance with Data Hiding](https://mlsanity.com/paper/2009.03015/) (2020)
- [Unbiased Watermark for Large Language Models](https://mlsanity.com/paper/2310.10669/) (2023)
- [Protecting Language Generation Models via Invisible Watermarking](https://mlsanity.com/paper/2302.03162/) (2023)
- [Three Bricks to Consolidate Watermarks for Large Language Models](https://mlsanity.com/paper/2308.00113/) (2023)
- [Adaptive Text Watermark for Large Language Models](https://mlsanity.com/paper/2401.13927/) (2024)
- [A Statistical Framework of Watermarks for Large Language Models: Pivot, Detection Efficiency and Optimal Rules](https://mlsanity.com/paper/2404.01245/) (2024)
- [A Nonparametric Approach for Multiple Change Point Analysis of Multivariate Data](https://mlsanity.com/paper/1306.4933/) (2013)
- [Towards Optimal Statistical Watermarking](https://mlsanity.com/paper/2312.07930/) (2023)
- [Permute-and-Flip: An optimally stable and watermarkable decoder for LLMs](https://mlsanity.com/paper/2402.05864/) (2024)
- [Towards Better Statistical Understanding of Watermarking LLMs](https://mlsanity.com/paper/2403.13027/) (2024)
- [Narrowest-Over-Threshold Detection of Multiple Change-points and Change-point-like Features](https://mlsanity.com/paper/1609.00293/) (2016)
- [Seeded Binary Segmentation: A general methodology for fast and optimal change point detection](https://mlsanity.com/paper/2002.06633/) (2020)
- [High-dimensional Change-point Detection Using Generalized Homogeneity Metrics](https://mlsanity.com/paper/2105.08976/) (2021)
- [Sequential Gradient Descent and Quasi-Newton's Method for Change-Point Analysis](https://mlsanity.com/paper/2210.12235/) (2022)
- [fastcpd: Fast Change Point Detection in R](https://mlsanity.com/paper/2404.05933/) (2024)

## Built on by (most cited first, 4 of 4)

- [Marking Code Without Breaking It: Code Watermarking for Detecting LLM-Generated Code](https://mlsanity.com/paper/2502.18851/) (2025)
- [Segmenting Human-LLM Co-authored Text via Change Point Detection](https://mlsanity.com/paper/2605.03723/) (2026)
- [Adaptive Testing for Segmenting Watermarked Texts From Language Models](https://mlsanity.com/paper/2511.06645/) (2025)
- [Anchor-ECC: Local Integrity Checking for Watermarked LLM Outputs via Error-Correcting Codes](https://mlsanity.com/paper/2609.38722/) (2026)

## Abstract (arXiv)

Watermarking is a technique that involves embedding nearly unnoticeable statistical signals within generated content to help trace its source. This work focuses on a scenario where an untrusted third-party user sends prompts to a trusted language model (LLM) provider, who then generates a text from their LLM with a watermark. This setup makes it possible for a detector to later identify the source of the text if the user publishes it. The user can modify the generated text by substitutions, insertions, or deletions. Our objective is to develop a statistical method to detect if a published text is LLM-generated from the perspective of a detector. We further propose a methodology to segment the published text into watermarked and non-watermarked sub-strings. The proposed approach is built upon randomization tests and change point detection techniques. We demonstrate that our method ensures Type I and Type II error control and can accurately identify watermarked sub-strings by finding the corresponding change point locations. To validate our technique, we apply it to texts generated by several language models with prompts extracted from Google's C4 dataset and obtain encouraging numerical results. We release all code publicly at https://github.com/doccstat/llm-watermark-cpd.