# Is Multilingual LLM Watermarking Truly Multilingual? Scaling Robustness to 100+ Languages via Back-Translation

arXiv: https://arxiv.org/abs/2510.18019 · HTML: https://mlsanity.com/paper/2510.18019/

## Key facts

- Cited by 1 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 15 of its 29 references that are in the map.
- First posted on arXiv 20 Oct 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, 15 of 15)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [GPyTorch: Blackbox Matrix-Matrix Gaussian Process Inference with GPU Acceleration](https://mlsanity.com/paper/1809.11165/) (2018)
- [Word Translation Without Parallel Data](https://mlsanity.com/paper/1710.04087/) (2017)
- [Robust Distortion-free Watermarks for Language Models](https://mlsanity.com/paper/2307.15593/) (2023)
- [Provable Robust Watermarking for AI-Generated Text](https://mlsanity.com/paper/2306.17439/) (2023)
- [Who Wrote this Code? Watermarking for Code Generation](https://mlsanity.com/paper/2305.15060/) (2023)
- [A Robust Semantics-based Watermark for Large Language Model against Paraphrasing](https://mlsanity.com/paper/2311.08721/) (2023)
- [An Entropy-based Text Watermarking Detection Method](https://mlsanity.com/paper/2403.13485/) (2024)
- [Watermarking Conditional Text Generation for AI Detection: Unveiling Challenges and a Semantic-Aware Watermark Remedy](https://mlsanity.com/paper/2307.13808/) (2023)
- [LLaMAX: Scaling Linguistic Horizons of LLM by Enhancing Translation Capabilities Beyond 100 Languages](https://mlsanity.com/paper/2407.05975/) (2024)
- [WatME: Towards Lossless Watermarking Through Lexical Redundancy](https://mlsanity.com/paper/2311.09832/) (2023)
- [URIEL+: Enhancing Linguistic Inclusion and Usability in a Typological and Multilingual Knowledge Base](https://mlsanity.com/paper/2409.18472/) (2024)
- [Lost in Overlap: Exploring Logit-based Watermark Collision in LLMs](https://mlsanity.com/paper/2403.10020/) (2024)
- [Evaluating the Robustness and Accuracy of Text Watermarking Under Real-World Cross-Lingual Manipulations](https://mlsanity.com/paper/2502.16699/) (2025)
- [Robustness Assessment and Enhancement of Text Watermarking for Google's SynthID](https://mlsanity.com/paper/2508.20228/) (2025)

## Built on by (most cited first, 1 of 1)

- [Linguistics-Aware Non-Distortionary LLM Watermarking](https://mlsanity.com/paper/2606.00613/) (2026)

## Abstract (arXiv)

Multilingual watermarking aims to make large language model (LLM) outputs traceable across languages, yet current methods still fall short. Despite claims of cross-lingual robustness, they are evaluated only on high-resource languages. We show that existing multilingual watermarking methods are not truly multilingual: they fail to remain robust under translation attacks in medium- and low-resource languages. We trace this failure to semantic clustering, which fails when the tokenizer vocabulary contains too few full-word tokens for a given language. To address this, we introduce STEAM, a detection method that uses Bayesian optimisation to search among 126 candidate languages for the back-translation that best recovers the watermark strength. It is compatible with any watermarking method, robust across different tokenizers and languages, non-invasive, and easily extendable to new languages. With average gains of +0.23 AUC and +37%p TPR@1%, STEAM provides a scalable approach toward fairer watermarking across the diversity of languages.