ReasMark: A Robust Watermark for Attributing LLM Reasoning Under Knowledge Distillation Attacks (2026.acl-long)
Copied to clipboard
Peizhuo Lv, Ruihua Zhou, Yunpeng Li, Ruigang Liang, Xingshuo Han, XiaoFeng Wang, Wei Dong, Yuling Liu
| Challenge: | Existing reasoning-enhanced large language models fail to provide reliable attribution of reasoning behavior once it is transferred through knowledge distillation. |
| Approach: | They propose to embed a reasoning-length gap in a model by querying a target domain and training a local student to imitate its outputs. |
| Outcome: | et al. show that ReasMark outperforms baselines while preserving task utility. |
Similar Papers
Can LLM Watermarks Robustly Prevent Unauthorized Knowledge Distillation? (2025.acl-long)
Copied to clipboard
| Challenge: | Large Language Model (LLM) watermarking is radioactive and enables the detection of watermarks inherited by student models when trained on the outputs of watermarked teacher models. |
| Approach: | They propose two types of watermark removal attacks that allow student models to perform untraceable knowledge distillation while avoiding watermark inheritance. |
| Outcome: | The proposed attacks eliminate inherited watermarks while maintaining knowledge transfer efficiency and low computational overhead. |
Protecting Language Models Against Unauthorized Distillation through Trace Rewriting (2026.acl-long)
Copied to clipboard
| Challenge: | Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more efficient student models. |
| Approach: | They propose methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) anti-distillation, or degrading the training usefulness of query responses; and (2) API watermarking, which embeds verifiable signatures in student models. |
| Outcome: | The proposed method achieves strong anti-distillation effect while maintaining or even improving teacher performance. |
DITTO: A Spoofing Attack Framework on Watermarked LLMs via Knowledge Distillation (2026.eacl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) generate coherent, human-like text at scale, but raises concerns about authenticity and trust. |
| Approach: | They propose a threat of watermark spoofing that allows a malicious model to generate text containing the authentic-looking watermark of a trusted, victim model. |
| Outcome: | The proposed attack repurposes watermark radioactivity from a discoverable trait into an attack vector and replicates it. |
Robust Data Watermarking in Language Models by Injecting Fictitious Knowledge (2025.findings-acl)
Copied to clipboard
| Challenge: | Data watermarking in language models injects traceable signals, such as specific token sequences or stylistic patterns, into copyrighted text, allowing copyright holders to track and verify training data ownership. |
| Approach: | They propose a data watermarking approach that injects coherent and plausible yet fictitious knowledge into training data using generated passages describing a fictious entity and its associated attributes. |
| Outcome: | The proposed method is designed to be memorized by the LLM, and that increasing their density, length, and diversity of attributes strengthens their memorization. |
PostMark: A Robust Blackbox Watermark for Large Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to detect LLM-generated text require access to the underlying LLM’s logits, which LLM providers are loath to share due to fears of model distillation. |
| Approach: | They develop a post-hoc watermarking procedure that inserts an input-dependent set of words into the text after the decoding process has completed. |
| Outcome: | The proposed method is more robust to paraphrasing attacks than existing methods. |
Distillation-Resistant Watermarking for Model Protection in NLP (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing protection methods such as watermarking only work for images but are not applicable to text. |
| Approach: | They propose a technique that injects watermarks into the victim’s prediction probability corresponding to a secret key and is able to detect such a key by probing a suspect model. |
| Outcome: | The proposed technique detects stealing suspects at 100% accuracy on four NLP tasks while the prior method fails on two. |
Rethinking LLM Watermark Detection in Black-Box Settings: A Non-Intrusive Third-Party Framework (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing secret-key schemes tightly couple detection with injection . this dependency creates a fundamental barrier for real-world governance . |
| Approach: | et al. introduce a black-box framework for non-intrusive, third-party watermark verification . they propose a proxy model to amplify watermark-relevant signals and complementary relative measurements . |
| Outcome: | a new framework decouples detection from injection and assesses alignment of query text with watermark distributions. |
LLMR: Knowledge Distillation with a Large Language Model-Induced Reward (2024.lrec-main)
Copied to clipboard
| Challenge: | Large language models have demonstrated remarkable performance in various NLP tasks, but are typically computationally expensive and difficult to be deployed in resource-constrained environments. |
| Approach: | They propose a knowledge distillation method based on a reward function induced from large language models. |
| Outcome: | The proposed method outperforms traditional methods on multiple datasets and tasks. |
SimMark: A Robust Sentence-Level Similarity-Based Watermarking Algorithm for Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Experimental results demonstrate that SimMark surpasses previous sentence-level watermarking techniques in robustness, sampling efficiency, and applicability across diverse domains. |
| Approach: | They propose a sentence-level watermarking algorithm that makes LLM outputs traceable without requiring access to model internals. |
| Outcome: | The proposed algorithm surpasses previous sentence-level watermarking techniques in robustness, sampling efficiency, and applicability across diverse domains while maintaining the text quality and fluency. |
XMark: Reliable Multi-Bit Watermarking for LLM-Generated Texts (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for embedding binary messages into LLM-generated text suffer from key limitations, such as a poor trade-off between text quality and decoding accuracy. |
| Approach: | They propose a method for embedding binary messages into Large Language Model (LLM)-generated text that uses a limited number of tokens to decode and recover the encoded message. |
| Outcome: | The proposed method significantly outperforms existing methods in multiple downstream tasks and will be made publicly available upon acceptance. |