SUV: Scalable Large Language Model Copyright Compliance with Regularized Selective Unlearning

2025-03-29

AI Summary

This work addresses the legal risks arising from large language models (LLMs) inadvertently generating copyright-infringing content. We propose Selective Unlearning via Verbalization (SUV), a selective forgetting framework that constructs a copyright-sensitive book dataset and innovatively integrates gradient projection with Fisher information matrix regularization into Direct Preference Optimization (DPO). SUV guides the model to replace verbatim memorization with semantically equivalent paraphrased outputs, enabling precise erasure of infringing content. Evaluated on 500 copyrighted books, SUV reduces exact verbatim reproduction rates significantly while incurring less than a 0.5% performance degradation on standard NLP benchmarks. This demonstrates SUV’s high efficacy, scalability, and practicality in preserving model capabilities while mitigating copyright violations.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Large Language Models (LLMs) have transformed natural language processing by learning from massive datasets, yet this rapid progress has also drawn legal scrutiny, as the ability to unintentionally generate copyrighted content has already prompted several prominent lawsuits. In this work, we introduce SUV (Selective Unlearning for Verbatim data), a selective unlearning framework designed to prevent LLM from memorizing copyrighted content while preserving its overall utility. In detail, the proposed method constructs a dataset that captures instances of copyrighted infringement cases by the targeted LLM. With the dataset, we unlearn the content from the LLM by means of Direct Preference Optimization (DPO), which replaces the verbatim copyrighted content with plausible and coherent alternatives. Since DPO may hinder the LLM's performance in other unrelated tasks, we integrate gradient projection and Fisher information regularization to mitigate the degradation. We validate our approach using a large-scale dataset of 500 famous books (predominantly copyrighted works) and demonstrate that SUV significantly reduces verbatim memorization with negligible impact on the performance on unrelated tasks. Extensive experiments on both our dataset and public benchmarks confirm the scalability and efficacy of our approach, offering a promising solution for mitigating copyright risks in real-world LLM applications.

Problem

• Prevent LLMs from memorizing copyrighted content
• Maintain model utility while unlearning copyrighted data
• Reduce copyright risks without degrading unrelated task performance

Innovation

• Selective unlearning framework for copyright compliance
• Uses Direct Preference Optimization for content replacement
• Integrates gradient projection and Fisher regularization