Uncolorable Examples: Preventing Unauthorized AI Colorization via Perception-Aware Chroma-Restrictive Perturbation

2025-10-09

AI Summary

To address copyright infringement risks arising from unauthorized AI-based colorization and resale of grayscale content, this paper proposes the first copyright defense paradigm specifically designed for grayscale images. Methodologically, we introduce the concept of “uncolorizable samples” and establish a four-dimensional evaluation framework—effectiveness, imperceptibility, transferability, and robustness. We design Perceptually Aware Chroma Perturbation (PAChroma), which incorporates Laplacian filtering to constrain perturbation distributions and employs diverse input transformations to enhance resilience against compression and cross-model generalization. Experiments on ImageNet and Danbooru demonstrate that our approach significantly degrades output quality of mainstream AI colorization models (PSNR reduction >8 dB) while preserving visual fidelity of the original grayscale images (SSIM >0.99). This work delivers the first verifiable, practical, and proactive defense mechanism for copyright protection in generative media.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

AI-based colorization has shown remarkable capability in generating realistic color images from grayscale inputs. However, it poses risks of copyright infringement -- for example, the unauthorized colorization and resale of monochrome manga and films. Despite these concerns, no effective method currently exists to prevent such misuse. To address this, we introduce the first defensive paradigm, Uncolorable Examples, which embed imperceptible perturbations into grayscale images to invalidate unauthorized colorization. To ensure real-world applicability, we establish four criteria: effectiveness, imperceptibility, transferability, and robustness. Our method, Perception-Aware Chroma-Restrictive Perturbation (PAChroma), generates Uncolorable Examples that meet these four criteria by optimizing imperceptible perturbations with a Laplacian filter to preserve perceptual quality, and applying diverse input transformations during optimization to enhance transferability across models and robustness against common post-processing (e.g., compression). Experiments on ImageNet and Danbooru datasets demonstrate that PAChroma effectively degrades colorization quality while maintaining the visual appearance. This work marks the first step toward protecting visual content from illegitimate AI colorization, paving the way for copyright-aware defenses in generative media.

Problem

• Prevent unauthorized AI colorization of grayscale images
• Embed imperceptible perturbations to disrupt colorization models
• Protect copyright of monochrome manga and films

Innovation

• Embed imperceptible perturbations to block unauthorized colorization
• Use Laplacian filter optimization for maintaining perceptual quality
• Apply input transformations for cross-model transferability and robustness