@inproceedings{li-etal-2026-agtao,
    title = "{AGT}{AO}: Robust and Stabilized {LLM} Unlearning via Adversarial Gating Training with Adaptive Orthogonality",
    author = "Li, Pengyu  and
      Zhang, Lingling  and
      Gao, Zhitao  and
      Wu, Yanrui  and
      Dong, Yuxuan  and
      Liu, Huan  and
      Wei, Bifan  and
      Liu, Jun",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.665/",
    pages = "13585--13600",
    ISBN = "979-8-89176-395-1",
    abstract = "While Large Language Models (LLMs) have achieved remarkable capabilities, they unintentionally memorize sensitive data, posing critical privacy and security risks.Machine unlearning is pivotal for mitigating these risks, yet existing paradigms face a fundamental dilemma: aggressive unlearning often induces catastrophic forgetting that degrades model utility, whereas conservative strategies risk superficial forgetting, leaving models vulnerable to adversarial recovery. To address this trade-off, we propose \textbf{AGT{AO}} (Adversarial Gating Training with Adaptive Orthogonality), a unified framework designed to reconcile robust erasure with utility preservation. Specifically, our approach introduces \textbf{Adaptive Orthogonality (AO)} to dynamically mitigate geometric gradient conflicts between forgetting and retention objectives, thereby minimizing unintended knowledge degradation. Concurrently, \textbf{Adversarial Gating Training (AGT)} formulates unlearning as a latent-space min-max game, employing a curriculum-based gating mechanism to simulate and counter internal recovery attempts. Extensive experiments demonstrate that AGT{AO} achieves a superior trade-off between unlearning efficacy (KUR $\approx$ 0.01) and model utility (MMLU 58.30).[Code is available at {\ensuremath{<}}https://anonymous.4open.science/r/AGT-unlearning{\ensuremath{>}}.]."
}