[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84359-en":3,"doc-seo-84359-105":30,"detail-sidebar-cat-0-en-105":83},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},84359,1099514068365,"Aurelia","https://ap-avatar.wpscdn.com/avatar/10000253d8d9f28188e?_k=1776742907772140068",8,"Research & Report","AutoAnchor: Stable Diffusion Unlearning Using Cross Attention as a Manifold Surrogate","Diffusion unlearning mitigates harmful or copyrighted content generation in text-to-image models. Existing approaches choose update directions via manually selected semantic anchors or via anchor-free alternatives using empty or target-irrelevant prompts. Anchor-based unlearning risks biased updates, while anchor-free updates can drift off the data manifold, causing unstable behavior. The work formalizes this instability under the manifold hypothesis and shows that missing manifold-proximal anchors induces normal-space drift. AutoAnchor synthesizes manifold-proximal anchors through a two-stage process using cross-attention consistency as a computationally efficient surrogate.","arXiv :2607 .08337v 1 [ cs .LG] 9 Jul 2026  \nAutoAnchor: Stable Diffusion Unlearning Using Cross-Attention as a Manifold Surrogate  \nSiyuan Wen 1 , Jiahao Zeng 1 , Ningning Ding 1 ∗  \n1Hong Kong University of Science and Technology (Guangzhou)  \n{swen211,[jzeng110}@connect.hkust-gz.edu.cn](jzeng110}@connect.hkust-gz.edu.cn), [ningningding@hkust-gz.edu.cn](ningningding@hkust-gz.edu.cn)  \nAbstract  \nDiffusion unlearning is essential for mitigating the generation of harmful or copyrighted content in text-to-image models. Current diffusion unlearning techniques determine the model update direction by either using alternatives of the target concept as an anchor or using empty prompts. The anchor-based method relies on manually and semantically-chosen anchors that risk biased unlearning, while the anchor-free method inherently suffers from unrobust unlearning due to unconstrained latent updates. In this work, we theoretically formalize such unstable diffusion unlearning issues under the manifold hypothesis and prove that lacking a manifold-proximal anchor inevitably induces significant normal-space drift that degrades unlearning performance. To achieve stable unlearning, we propose AutoAnchor, a two-stage framework that automatically synthesizes manifold-proximal anchors. However, direct geometric manifold optimization is computationally intractable. To address this challenge, AutoAnchor introduces a novel cross-attention consistency loss which serves as a highly efficient surrogate of manifold proximity.  \nExperimental results demonstrate that AutoAnchor effectively achieves robust and unbiased unlearning across various state-of-the-art baselines, significantly improving targeted concept removal (by up to 31.04% in CLIP score) and non-target utility (by up to 4.18% in CLIP score) . Moreover, AutoAnchor can also be easily integrated into existing diffusion unlearning methods to enhance their unlearning performance (by 6.30% for concept removal and 6.65% for utility on average) .  \n1 Introduction  \nText-to-Image diffusion models have achieved remarkable success in generating high-fidelity images conditioned on input text prompts. However, their reliance on massive and uncurated web data introduces significant risks, e.g., generating harmful content or copyright infringement. To mitigate these risks, diffusion unlearning has emerged as a crucial technique. The goal of diffusion unlearning is to erase a diffusion model’s ability to generate specific target concepts, which is usually realized by tuning the model parameters towards certain update directions.  \nAchieving effective erasure while preserving model utility is hard for diffusion unlearning. This is because the modern diffusion models operate in a high-dimensional latent space, but the valid and realistic images reside on a low-dimensional data manifold within that. Without a carefully designed update direction, diffusion unlearning can easily push the model off the data manifold when erasing target concepts, thereby degrading the erasure effectiveness and damaging the model’s utility.  \nRegarding how to determine the update directions, current diffusion unlearning methods are categorized into anchor-free unlearning and anchor-based unlearning. Anchor-free methods (e.g., [13, 15, 29]) utilize target-irrelevant directions for unlearning, simply pushing the genera  \ntion away from the target concepts. For example, they tune the models toward valid but unconditional ∗Corresponding author.  \nPreprint.  \nTarget Concept  \nManual Anchor AutoAnchor  \nSubmanifold of Tangent Concept Submanifold of Manual Anchor Submanifold of AutoAnchor  \nData Manifold  \nGeodesic on the Manifold Surface  \nUpdate Directions no anchor/ biased anchor/ AutoAnchor  \nGray Plane:  \nManifold’s Tagent Space  \nColorful Arrow:  \nUnlearning Directions  \nGray Arrow:  \nNormal Components  \nBlack Arrow:  \nTangent Components  \nEach point on the manifold surface corresponds to a valid image in the real world.  \nFor  , the","cbCailo6vfyQCgQ7","https://ap.wps.com/l/cbCailo6vfyQCgQ7","pdf",9602164,4,1,28,"English","en",105,"# Abstract\n# Introduction","[{\"question\":\"How does AutoAnchor improve stable and unbiased unlearning?\",\"answer\":\"AutoAnchor automatically synthesizes manifold-proximal anchors using a two-stage framework. It avoids expensive geometric manifold optimization by introducing a cross-attention consistency loss as a surrogate for manifold proximity, improving targeted concept removal while preserving non-target utility.\"}]",1784195071,71,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":78,"head_meta":80,"extra_data":82,"updated_unix":28},"autoanchor-stable-diffusion-unlearning-using-cross-attention-as-a-manifold-surrogate","",{"@graph":36,"@context":77},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":20},"https://docshare.wps.com/document/autoanchor-stable-diffusion-unlearning-using-cross-attention-as-a-manifold-surrogate/84359/",{"url":52,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-26","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"How does AutoAnchor improve stable and unbiased unlearning?","Question",{"text":75,"@type":76},"AutoAnchor automatically synthesizes manifold-proximal anchors using a two-stage framework. It avoids expensive geometric manifold optimization by introducing a cross-attention consistency loss as a surrogate for manifold proximity, improving targeted concept removal while preserving non-target utility.","Answer","https://schema.org",{"og:url":52,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":46,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":46,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":46,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]