arXiv cs.AI· Junkai Lin, Ge Xu, Zheng Xu, Junkai Chen, Xiangyuan Lan·· 3 天前AI 评分38
LLM 智能体能否真正遗忘？AUBench 基准探索智能体机器遗忘
Can Agents Truly Forget? Explore Machine Unlearning for LLM Agents
AI 导读
研究者提出智能体遗忘问题，即需从 LLM 智能体的参数知识、持久记忆和工具调用三条信息通道中移除隐私信息，仅靠参数级遗忘并不足够。为此推出 Agent Unlearning Benchmark（AUBench），将 TOFU 的虚构作者设定扩展至带持久记忆、离线网页快照和工具可访问文档的智能体环境。评测显示模型级遗忘与系统级智能体遗忘之间存在明显差距。
正文 · 原文
Abstract:Large language model (LLM) agents have shown strong capabilities in interactive problem solving, tool use, and long-horizon task execution. However, they may also acquire sensitive or private knowledge during training and expose it during deployment. Retraining a model from scratch to remove such knowledge is prohibitively costly, making machine unlearning a more practical approach. Machine unlearning aims to remove targeted undesirable knowledge from trained models without retraining from scratch while preserving general utility. When language models are deployed as agents, parameter-level unlearning becomes insufficient: forgotten information may still persist in memory artifacts or be reacquired via tool calls. In this paper, we formulate the problem of agent unlearning as removing private information from an LLM-based agent across three information channels: parametric knowledge, persistent memory, and tool-call acquisition. To study this problem, we introduce we introduce Agent Unlearning Benchmark (AUBench), a controlled benchmark that extends the fictitious-author setting of TOFU into an agent environment with persistent memory, an offline web snapshot, tool-accessible documents, and provenance-tracked interaction trajectories. By evaluating existing parameter-level and memory-level unlearning methods in this setting, AUBench reveals the gap between model-level forgetting and system-level agent unlearning, and provides a foundation for developing unlearning methods that operate across the full lifecycle of LLM agents.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.17659 [cs.AI] |
| (or arXiv:2610.17659v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.17659 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Junkai Lin [view email]
[v1]
Tue, 13 Oct 2026 03:37:30 UTC (548 KB)
来源：arXiv cs.AI · arxiv.org