mlsanityData to 7 Oct 2026

SimMark: A Robust Sentence-Level Similarity-Based Watermarking Algorithm for Large Language Models

arXiv 2502.02787 ↗First posted 5 Feb 2025cs.CL cs.CR cs.CYcited by 13

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

Abstract (arXiv)

The widespread adoption of large language models (LLMs) necessitates reliable methods to detect LLM-generated text. We introduce SimMark, a robust sentence-level watermarking algorithm that makes LLMs' outputs traceable without requiring access to model internals, making it compatible with both open and API-based LLMs. By leveraging the similarity of semantic sentence embeddings combined with rejection sampling to embed detectable statistical patterns imperceptible to humans, and employing a soft counting mechanism, SimMark achieves robustness against paraphrasing attacks. Experimental results demonstrate that SimMark sets a new benchmark for robust watermarking of LLM-generated content, surpassing prior sentence-level watermarking techniques in robustness, sampling efficiency, and applicability across diverse domains, all while maintaining the text quality and fluency.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security. Builds on 22 of its 45 references that are in the map. First posted on arXiv 5 Feb 2025.

Builds on

22 of 45 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024MarkLLM: An Open-Source Toolkit for LLM Watermarking16 May
  2. 2024k-SemStamp: A Clustering-Based Semantic Watermark for Detection of Machine-Generated Text17 Feb
  3. 2023Watermarking Conditional Text Generation for AI Detection: Unveiling Challenges and a Semantic-Aware Watermark Remedy25 Jul
  4. 2023Provable Robust Watermarking for AI-Generated Text30 Jun
  5. 2023Watermarking Text Generated by Black-Box Language Models14 May
  6. 2023Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense23 Mar
  7. 2023Can AI-Generated Text be Reliably Detected?17 Mar

Other references · 15

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,767
  3. 2019BERTScore: Evaluating Text Generation with BERTcited by 5,462
  4. 2020DeBERTa: Decoding-enhanced BERT with Disentangled Attentioncited by 2,157
  5. 2022Large Language Models Encode Clinical Knowledgecited by 1,807
  6. 2019PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarizationcited by 1,100
  7. 2021Ethical and social risks of harm from Language Modelscited by 957
  8. 2022One Embedder, Any Task: Instruction-Finetuned Text Embeddingscited by 329
  9. 2021BookSum: A Collection of Datasets for Long-form Narrative Summarizationcited by 184
  10. 2018Generating Informative and Diverse Conversational Responses via Adversarial Information Maximizationcited by 114
  11. 2018Abstractive Summarization of Reddit Posts with Multi-level Memory Networkscited by 102
  12. 2024Logits of API-Protected LLMs Leak Proprietary Informationcited by 38
3 more
  1. 2023How Reliable Are AI-Generated-Text Detectors? An Assessment Framework Using Evasive Soft Promptscited by 23
  2. 2022Measuring and Improving Semantic Diversity of Dialogue Generationcited by 16
  3. 2024Empowering Air Travelers: A Chatbot for Canadian Air Passenger Rightscited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 5; 2026: 8.

13 papers.

Sort:

By area:

  1. 2025PMark: Towards Robust and Distortion-free Semantic-level Watermarking with Channel Constraintscited by 14
  2. 2025Majority Bit-Aware Watermarking For Large Language Modelscited by 3
  3. 2026Distilling the Thought, Watermarking the Answer: A Principle Semantic Guided Watermark for Large Reasoning Modelscited by 3
  4. 2026A Visual Semantic Adaptive Watermark grounded by Prefix-Tuning for Large Vision-Language Modelcited by 3
  5. 2025VIDSTAMP: A Temporally-Aware Watermark for Ownership and Integrity in Video Diffusion Modelscited by 1
  6. 2026SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustnesscited by 1
  7. 2026Global Sketch-Based Watermarking for Diffusion Language Modelscited by 1
  8. 2026DenMark: Robust Semantic Watermarking for Diffusion Language Modelscited by 1
  9. 2025LLM Watermark Evasion via Bias Inversioncited by 0
  10. 2025WaterSearch: Exploring Seed Pooling for Improving the Quality-Detectability Trade-off in LLM Watermarkingcited by 0
  11. 2026LLM Watermarking as Big Data Provenance: A Deployment-Oriented Systematizationcited by 0
  12. 2026Semantic Watermarking with Order-Robust Detection over Sub-sentence Unitscited by 0
1 more
  1. 2026Beyond Semantic Narrowing: Robust and Efficient LLM Watermarking with Hamming Neighborhoodscited by 0