Unlearning in LLMs: Methods, Evaluation, and Open Challenges
It addresses the need for responsible AI deployment by synthesizing progress in machine unlearning for LLMs, but it is incremental as a survey rather than presenting new methods.
This survey paper tackles the problem of selectively removing knowledge from large language models (LLMs) to address privacy, copyright, security, and bias concerns, by providing a structured overview of unlearning methods, evaluation benchmarks, and open challenges.
Large language models (LLMs) have achieved remarkable success across natural language processing tasks, yet their widespread deployment raises pressing concerns around privacy, copyright, security, and bias. Machine unlearning has emerged as a promising paradigm for selectively removing knowledge or data from trained models without full retraining. In this survey, we provide a structured overview of unlearning methods for LLMs, categorizing existing approaches into data-centric, parameter-centric, architecture-centric, hybrid, and other strategies. We also review the evaluation ecosystem, including benchmarks, metrics, and datasets designed to measure forgetting effectiveness, knowledge retention, and robustness. Finally, we outline key challenges and open problems, such as scalable efficiency, formal guarantees, cross-language and multimodal unlearning, and robustness against adversarial relearning. By synthesizing current progress and highlighting open directions, this paper aims to serve as a roadmap for developing reliable and responsible unlearning techniques in large language models.