A Survey on Generative Model Unlearning: Fundamentals, Taxonomy, Evaluation, and Future Direction

2025-07-26

AI Summary

The rapid advancement of generative models has intensified privacy risks, yet existing machine unlearning research lacks a unified framework—leading to inconsistent forgetting objectives, methodological strategies, and evaluation protocols, thereby impeding fair comparative analysis. To address this, we propose GenMU, the first systematic analytical framework for unlearning in generative models. GenMU comprehensively categorizes forgetting objectives, organizes methodological strategies along a coherent spectrum, and defines multidimensional evaluation metrics, while uncovering intrinsic connections to model editing and RLHF. Through rigorous literature synthesis and conceptual abstraction, we establish a taxonomy that ensures both theoretical consistency and practical applicability. Concurrently, we release an open-source repository featuring standardized benchmarks and reproducible tooling, significantly enhancing method comparability and field-wide standardization. GenMU thus lays a foundational groundwork for privacy-preserving generative AI.

Technology Category

• Machine Learning: Privacy
• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

With the rapid advancement of generative models, associated privacy concerns have attracted growing attention. To address this, researchers have begun adapting machine unlearning techniques from traditional classification models to generative settings. Although notable progress has been made in this area, a unified framework for systematically organizing and integrating existing work is still lacking. The substantial differences among current studies in terms of unlearning objectives and evaluation protocols hinder the objective and fair comparison of various approaches. While some studies focus on specific types of generative models, they often overlook the commonalities and systematic characteristics inherent in Generative Model Unlearning (GenMU). To bridge this gap, we provide a comprehensive review of current research on GenMU and propose a unified analytical framework for categorizing unlearning objectives, methodological strategies, and evaluation metrics. In addition, we explore the connections between GenMU and related techniques, including model editing, reinforcement learning from human feedback, and controllable generation. We further highlight the potential practical value of unlearning techniques in real-world applications. Finally, we identify key challenges and outline future research directions aimed at laying a solid foundation for further advancements in this field. We consistently maintain the related open-source materials at https://github.com/caxLee/Generative-model-unlearning-survey.

Problem

• Lack unified framework for generative model unlearning
• Diverse unlearning objectives hinder fair approach comparison
• Need systematic review of GenMU techniques and applications

Innovation

• Adapting machine unlearning to generative models
• Proposing unified framework for GenMU categorization
• Exploring connections with model editing techniques