KinGuard: Hierarchical Kinship-Aware Fingerprinting to Defend Against Large Language Model Stealing

2026-01-19

AI Summary

This work addresses the inherent tension between stealth and robustness in existing backdoor-based fingerprinting methods for large language models, which often rely on high-perplexity triggers that induce detectable statistical anomalies. To overcome this limitation, the authors propose a novel knowledge-embedding watermarking mechanism that eschews explicit surface-level triggers in favor of semantic memory. Specifically, they construct a private knowledge base comprising structured kinship narratives, internalize this knowledge into the model via incremental pretraining, and verify ownership using concept-understanding probes. By embedding watermarks as implicit semantic associations rather than overt backdoors, the method achieves exceptional robustness, stealth, and efficacy under diverse adversarial conditions—including fine-tuning, input perturbations, and model merging—thereby significantly enhancing the security and practicality of model copyright protection.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Protecting the intellectual property of large language models requires robust ownership verification. Conventional backdoor fingerprinting, however, is flawed by a stealth-robustness paradox: to be robust, these methods force models to memorize fixed responses to high-perplexity triggers, but this targeted overfitting creates detectable statistical artifacts. We resolve this paradox with KinGuard, a framework that embeds a private knowledge corpus built on structured kinship narratives. Instead of memorizing superficial triggers, the model internalizes this knowledge via incremental pre-training, and ownership is verified by probing its conceptual understanding. Extensive experiments demonstrate KinGuard's superior effectiveness, stealth, and resilience against a battery of attacks including fine-tuning, input perturbation, and model merging. Our work establishes knowledge-based embedding as a practical and secure paradigm for model fingerprinting.

Problem

• model stealing
• fingerprinting
• intellectual property protection
• stealth-robustness paradox
• large language models

Innovation

• knowledge-based fingerprinting
• kinship-aware embedding
• incremental pre-training
• ownership verification
• stealth-robustness paradox