mlsanityData to 14 Oct 2026

Can Watermarking Techniques Help Prevent LLM Model Stealing?

arXiv 2607.10794 ↗First posted 12 Jul 2026cs.CRcited by 0

On the map: LLM safety and security → Watermarking generated content and models

Abstract (arXiv)

Model stealing attacks have recently been introduced, enabling the extraction of precise information from black-box commercial language models. In this work, we propose defense methods against a recent attack of citecarlini2024stealing and extensions for extracting the hidden layer dimension of production language models. Our methods are inspired by watermarking techniques that perturb the logits layer of these models to prevent such attacks. We provide empirical experiments demonstrating the effectiveness of the proposed defense versus model quality degradation across various configurations, and propose an effective defense against such attacks while preserving model utility.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security. Builds on 22 of its 38 references that are in the map. First posted on arXiv 12 Jul 2026.

Builds on

22 of 38 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024Can Watermarking Large Language Models Prevent Copyrighted Text Generation and Hide Training Data?24 Jul
  2. 2024Waterfall: Framework for Robust and Scalable Text Watermarking and Provenance for LLMs5 Jul
  3. 2023REMARK-LLM: A Robust and Efficient Watermarking Framework for Generative Large Language Models18 Oct
  4. 2023Watermarking Text Generated by Black-Box Language Models14 May
  5. 2023DeepTextMark: A Deep Learning-Driven Text Watermarking Approach for Identifying Large Language Model Generated Text9 May
  6. 2023Protecting Language Generation Models via Invisible Watermarking6 Feb
  7. 2023A Watermark for Large Language Models24 Jan

Other references · 15

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  5. 2020Knowledge Distillation: A Surveycited by 1,603
  6. 2016Stealing Machine Learning Models via Prediction APIscited by 619
  7. 2024A Survey on Knowledge Distillation of Large Language Modelscited by 284
  8. 2019High Accuracy and High Fidelity Extraction of Neural Networkscited by 214
  9. 2024Stealing Part of a Production Language Modelcited by 159
  10. 2020Adversarial Watermarking Transformer: Towards Tracing Text Provenance with Data Hidingcited by 130
  11. 2018Model Reconstruction from Model Explanationscited by 86
  12. 2019Reverse-Engineering Deep ReLU Networkscited by 83
3 more
  1. 2020Cryptanalytic Extraction of Neural Network Modelscited by 68
  2. 2023Language Model Inversioncited by 68
  3. 2024Polynomial Time Cryptanalytic Extraction of Deep Neural Networks in the Hard-Label Settingcited by 9

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025δ-STEAL: LLM Stealing Attack with Local Differential Privacy
  2. 2025SoK: Are Watermarks in LLMs Ready for Deployment?
  3. 2024Watermark Stealing in Large Language Models
  4. 2024Learnable Linguistic Watermarks for Tracing Model Extraction Attacks on Large Language Models
  5. 2024A Watermark for Black-Box Language Models
  6. 2025Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends
  7. 2024Provably Robust Watermarks for Open-Source Language Models
  8. 2024ModelShield: Adaptive and Robust Watermark against Model Extraction Attack