More Haste, Less Speed: Weaker Single-Layer Watermark Improves Distortion-Free Watermark Ensembles
cs.CR, cs.CL
Submitted: 2026-02-12
Updated: 2026-08-28
Terminology
Sources
- The Llama 3 Herd of Models
- BiMark: Unbiased Multilayer Watermarking for Large Language Models
- Generative Language Models and Automated Influence Operations: Emerging Threats and Potential Mitigations
- Unbiased Watermark for Large Language Models
- Mistral 7B
- A Watermark for Order-Agnostic Language Models
- Robust Distortion-free Watermarks for Language Models
- De-mark: Watermark Removal in Large Language Models
- Adaptive Text Watermark for Large Language Models
- Improved Unbiased Watermark for Large Language Models
- Watermarking Low-entropy Generation for Large Language Models: An Unbiased and Low-risk Method
- Undetectable Watermarks for Language Models
- Mark My Words: Analyzing and Evaluating Language Model Watermarks
- Can AI-Generated Text be Reliably Detected?
- WaterBench: Towards Holistic Evaluation of Watermarks for Large Language Models
- Ethical and social risks of harm from Language Models
- Distortion-free Watermarks are not Truly Distortion-free under Watermark Key Collisions
- An Ensemble Framework for Unbiased Language Model Watermarking
- Analyzing and Evaluating Unbiased Language Model Watermark
- A Resilient and Accessible Distribution-Preserving Watermark for Large Language Models
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs