Copyright Infringement Detection in Text-to-Image Diffusion Models via Differential Privacy

2025-09-26

AI Summary

Text-to-image diffusion models (e.g., Stable Diffusion) may inadvertently memorize and reproduce copyrighted content without authorization, yet existing infringement detection methods suffer from poor robustness and weak theoretical foundations. Method: This paper proposes DPM, the first differential privacy–based infringement detection framework. DPM simulates model learning and forgetting dynamics, introduces a conditional sensitivity metric to quantify generative dependence on training data, and leverages statistical analysis over orthogonal prompt distributions—enabling reliable detection without access to the original training data or ground-truth prompts. Contributions/Results: (1) First systematic integration of differential privacy theory into copyright detection; (2) A principled, interpretable framework—D-Plus-Minus (DPM); (3) The first dedicated benchmark, CIDD. Experiments demonstrate that DPM substantially improves detection robustness, theoretical rigor, and practical applicability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Privacy
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Data transparency and provenance
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

The widespread deployment of large vision models such as Stable Diffusion raises significant legal and ethical concerns, as these models can memorize and reproduce copyrighted content without authorization. Existing detection approaches often lack robustness and fail to provide rigorous theoretical underpinnings. To address these gaps, we formalize the concept of copyright infringement and its detection from the perspective of Differential Privacy (DP), and introduce the conditional sensitivity metric, a concept analogous to sensitivity in DP, that quantifies the deviation in a diffusion model's output caused by the inclusion or exclusion of a specific training data point. To operationalize this metric, we propose D-Plus-Minus (DPM), a novel post-hoc detection framework that identifies copyright infringement in text-to-image diffusion models. Specifically, DPM simulates inclusion and exclusion processes by fine-tuning models in two opposing directions: learning or unlearning. Besides, to disentangle concept-specific influence from the global parameter shifts induced by fine-tuning, DPM computes confidence scores over orthogonal prompt distributions using statistical metrics. Moreover, to facilitate standardized benchmarking, we also construct the Copyright Infringement Detection Dataset (CIDD), a comprehensive resource for evaluating detection across diverse categories. Our results demonstrate that DPM reliably detects infringement content without requiring access to the original training dataset or text prompts, offering an interpretable and practical solution for safeguarding intellectual property in the era of generative AI.

Problem

• Detects copyright infringement in text-to-image diffusion models
• Quantifies training data influence via differential privacy metrics
• Provides robust detection without original training data access

Innovation

• Differential Privacy framework detects copyright infringement
• DPM fine-tunes models in opposing learning directions
• Orthogonal prompt distributions compute confidence scores