DRAGON: Guard LLM Unlearning in Context via Negative Detection and Reasoning

TL;DR

DRAGON is a reasoning-based framework that enhances the unlearning of private or harmful data in LLMs without needing retain data, using in-context detection and reasoning to ensure safe model behavior.

Contribution

It introduces a novel, instruction-based unlearning method that does not require access to retain data and employs in-context reasoning for effective unlearning in practical scenarios.

Findings

DRAGON achieves strong unlearning performance across multiple tasks.

It demonstrates scalability and effectiveness without retraining the base model.

The framework introduces new metrics for evaluating unlearning performance.

Abstract

Unlearning in Large Language Models (LLMs) is crucial for protecting private data and removing harmful knowledge. Most existing approaches rely on fine-tuning to balance unlearning efficiency with general language capabilities. However, these methods typically require training or access to retain data, which is often unavailable in real world scenarios. Although these methods can perform well when both forget and retain data are available, few works have demonstrated equivalent capability in more practical, data-limited scenarios. To overcome these limitations, we propose Detect-Reasoning Augmented GeneratiON (DRAGON), a systematic, reasoning-based framework that utilizes in-context chain-of-thought (CoT) instructions to guard deployed LLMs before inference. Instead of modifying the base model, DRAGON leverages the inherent instruction-following ability of LLMs and introduces a lightweight detection module to identify forget-worthy prompts without any retain data. These are then routed through a dedicated CoT guard model to enforce safe and accurate in-context intervention. To robustly evaluate unlearning performance, we introduce novel metrics for unlearning performance and the continual unlearning setting. Extensive experiments across three representative unlearning tasks validate the effectiveness of DRAGON, demonstrating its strong unlearning capability, scalability, and applicability in practical scenarios.