SELF: A Robust Singular Value and Eigenvalue Approach for LLM Fingerprinting

2025/12/03 by H. Zhang, Yue Zheng, Zhang, Hanxiu +1
Computer Science · #Advanced Graph Neural Networks #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #Authorship Attribution and Profiling #Computation and Language (cs.CL) #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · pdf · doi:10.48550/arxiv.2512.03620

openalex publication_date 2025/12/03 · openalex created_date 2025/12/05 · openalex updated_date 2026/07/28

Abstract

The protection of Intellectual Property (IP) in Large Language Models (LLMs) represents a critical challenge in contemporary AI research. While fingerprinting techniques have emerged as a fundamental mechanism for detecting unauthorized model usage, existing methods -- whether behavior-based or structural -- suffer from vulnerabilities such as false claim attacks or susceptible to weight manipulations. To overcome these limitations, we propose SELF, a novel intrinsic weight-based fingerprinting scheme that eliminates dependency on input and inherently resists false claims. SELF achieves robust IP protection through two key innovations: 1) unique, scalable and transformation-invariant fingerprint extraction via singular value and eigenvalue decomposition of LLM attention weights, and 2) effective neural network-based fingerprint similarity comparison based on few-shot learning and data augmentation. Experimental results demonstrate SELF maintains high IP infringement detection accuracy while showing strong robustness against various downstream modifications, including quantization, pruning, and fine-tuning attacks. Our code is available at https://github.com/HanxiuZhang/SELF_v2.

Citations

• FIT-Print: Towards False-claim-resistant Model Ownership Verification via Targeted Fingerprint
• REEF: Representation Encoding Fingerprints for Large Language Models
• Scalable watermarking for identifying large language model outputs
• PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference
• LLM-Based Test-Driven Interactive Code Generation: User Study and Empirical Evaluation
• TRAP: Targeted Random Adversarial Prompt Honeypot for Black-Box Identification
• SliceGPT: Compress Large Language Models by Deleting Rows and Columns
• Instructional Fingerprinting of Large Language Models
• HuRef: HUman-REadable Fingerprint for Large Language Models
• Turning Your Strength into Watermark: Watermarking Large Language Model via Knowledge Injection
• Can LLMs Keep a Secret? Testing Privacy Implications of Language Models via Contextual Integrity Theory
• REMARK-LLM: A Robust and Efficient Watermarking Framework for Generative Large Language Models
• Watermarking LLMs with Weight Quantization
• A Semantic Invariant Robust Watermark for Large Language Models
• False Claims against Model Ownership Resolution
• Can AI-Generated Text be Reliably Detected?
• A Watermark for Large Language Models
• ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection
• TruthfulQA: Measuring How Models Mimic Human Falsehoods
• Regularizing and Optimizing LSTM Language Models
• Attention Is All You Need
• Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
• Pointer Sentinel Mixture Models
• Deep Residual Learning for Image Recognition
• Norms and exclusion theorems

Cited by

• Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends

Related

• iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification
• A Fingerprint for Large Language Models
• DuFFin: A Dual-Level Fingerprinting Framework for LLMs IP Protection
• Hey, That's My Model! Introducing Chain & Hash, An LLM Fingerprinting Technique
• ProFLingo: A Fingerprinting-based Intellectual Property Protection Scheme for Large Language Models