A survey of text watermarking in the era of large language models

A Liu, L Pan, Y Lu, J Li, X Hu, X Zhang, L Wen… - ACM Computing …, 2024 - dl.acm.org
Text watermarking algorithms are crucial for protecting the copyright of textual content.
Historically, their capabilities and application scenarios were limited. However, recent …

A watermark for large language models

J Kirchenbauer, J Geiping, Y Wen… - International …, 2023 - proceedings.mlr.press
Potential harms of large language models can be mitigated by watermarking model output,
ie, embedding signals into generated text that are invisible to humans but algorithmically …

Can AI-generated text be reliably detected?

VS Sadasivan, A Kumar, S Balasubramanian… - arXiv preprint arXiv …, 2023 - arxiv.org
The unregulated use of LLMs can potentially lead to malicious consequences such as
plagiarism, generating fake news, spamming, etc. Therefore, reliable detection of AI …

The science of detecting llm-generated text

R Tang, YN Chuang, X Hu - Communications of the ACM, 2024 - dl.acm.org
ACM: Digital Library: Communications of the ACM ACM Digital Library Communications of the
ACM Volume 67, Number 4 (2024), Pages 50-59 The Science of Detecting LLM-Generated Text …

Undetectable watermarks for language models

M Christ, S Gunn, O Zamir - The Thirty Seventh Annual …, 2024 - proceedings.mlr.press
Recent advances in the capabilities of large language models such as GPT-4 have spurred
increasing concern about our ability to detect AI-generated text. Prior works have suggested …

A review of text watermarking: theory, methods, and applications

NS Kamaruddin, A Kamsin, LY Por, H Rahman - IEEE Access, 2018 - ieeexplore.ieee.org
During the recent years, the issue of preserving the integrity of digital text has become a
focus of interest in the transmission of online content on the Internet. Watermarking has a …

Identifying and mitigating the security risks of generative ai

C Barrett, B Boyd, E Bursztein, N Carlini… - … and Trends® in …, 2023 - nowpublishers.com
Every major technical invention resurfaces the dual-use dilemma—the new technology has
the potential to be used for good as well as for harm. Generative AI (GenAI) techniques, such …

Robust distortion-free watermarks for language models

R Kuditipudi, J Thickstun, T Hashimoto… - arXiv preprint arXiv …, 2023 - arxiv.org
We propose a methodology for planting watermarks in text from an autoregressive language
model that are robust to perturbations without changing the distribution over text up to a …

Unbiased watermark for large language models

Z Hu, L Chen, X Wu, Y Wu, H Zhang… - arXiv preprint arXiv …, 2023 - arxiv.org
The recent advancements in large language models (LLMs) have sparked a growing
apprehension regarding the potential misuse. One approach to mitigating this risk is to …

Provable robust watermarking for ai-generated text

X Zhao, P Ananth, L Li, YX Wang - arXiv preprint arXiv:2306.17439, 2023 - arxiv.org
As AI-generated text increasingly resembles human-written content, the ability to detect
machine-generated text becomes crucial. To address this challenge, we present …