[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83952-en":3,"doc-seo-83952-105":30,"detail-sidebar-cat-0-en-105":91},{"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},83952,687197207639,"Asher","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Position Preventing AI-Generated CSAM Necessitates New Approaches to AI Safety","Modern artificial intelligence (AI) systems introduce severe, novel threats to child safety by enabling the creation of AI-generated child sexual abuse material (AIG-CSAM), supporting child sexual exploitation, and lowering barriers to harm. The paper argues that child protection against AI-facilitated abuse requires new AI-safety approaches because conventional techniques rely on assumptions incompatible with ethical and legal constraints. It analyzes technical challenges in dataset auditing, red teaming, and prevention via fine-tuning, then enumerates 15 lifecycle open problems and offers guidance for researchers, developers, and policymakers to translate safety principles into enforceable safeguards.","Position: Preventing AI-Generated CSAM Necessitates New Approaches to AI Safety  \nNeil Kale * 1 Rebecca Portnoff * 2 Pratiksha Thaker 1 Michael Simpson 2 Robertson Wang 2 Kevin Kuo 1 Chhavi Yadav 1 Virginia Smith 1  \narXiv :2607 .05407v 1 [ cs .CY] 9 Jun 2026  \nAbstract  \nModern artificial intelligence (AI) systems present profound new risks to child safety. AI is increasingly being misused to create AIgenerated child sexual abuse material, facilitate child sexual exploitation, and reduce barriers to harm. In this paper, we argue that protecting children from AI-facilitated sexual abuse requires new approaches to AI safety. Existing safety techniques assume data accessibility, transparency, and evaluation practices that are incompatible with the ethical and legal constraints surrounding child sexual abuse material. We examine how these constraints create new technical challenges, such as limitations on dataset auditing, red teaming, and fine-tuning prevention.  \nIn turn, we outline 15 open problems in online child sexual exploitation and abuse across the AI development lifecycle, from dataset curation and model design to deployment and long-term maintenance. We propose targeted recommendations for researchers, developers, and policymakers to bridge the gap between theoretical AI safety and the realities of child protection. Our work aims to reframe preventing AI-facilitated child sexual abuse as a central, safety-critical dimension for AI research, motivating work that translates responsible AI principles into concrete safeguards against the exploitation of children.  \n1. Introduction  \nArtificial intelligence (AI) systems have achieved remarkable capabilities in content creation. However, these same capabilities increasingly pose risks to children, including facilitating child sexual abuse and exploitation (CSAE), particularly via the development of AI-generated child sex-  \nThe first two authors contributed equally and may list their names interchangeably. 1Carnegie Mellon University 2Thorn. Correspondence to: Neil Kale \u003C[nkale@cs.cmu.edu](nkale@cs.cmu.edu) >, Rebecca Portnoff \u003C[rebecca@wearethorn.org](rebecca@wearethorn.org) >.  \nProceedings of the 43 rd International Conference on Machine Learning, Seoul, South Korea. PMLR 306, 2026 . Copyright 2026 by the author(s) .  \nual abuse material (AIG-CSAM) . This misuse represents a critical challenge at the intersection of AI safety, child protection, and technical system design—one that most existing AI safety frameworks inadequately address (IWF, 2023; 2024; Paltieli & Freud, 2023; Thiel et al., 2023) .  \nThe scope and severity of this problem are substantial. The National Center for Missing and Exploited Children (NCMEC) received more than 440,000 reports of AIgenerated material related to CSAE in the first half of 2025 alone (Davis, 2025), compared to 7,000 reports in 2023- 2024, and the Internet Watch Foundation (IWF) observed a 400% increase in AIG-CSAM reports since 2024 (IWF, 2025; 2024). A recent study found that [13% of U.S. sex](13% of U.S. sex)ual extortion victims reported that the perpetrator used AI to create blackmail material (Thorn, 2025), and a 2025 survey of Australian young people found that 41% of sexual extortion victims reported that the material used to blackmail them had been digitally manipulated (Wolbers et al., 2025) . In the U.S., 1 in 17 teenagers aged 13–17 report having been victimized by deepfake nude images (Thorn, 2024a) . AIG-CSAM has thus been recognized by the AI safety community as a critical area for intervention, with immediate, quantifiable marginal risk (Kapoor et al., 2024) .  \nHowever, despite the known risks of AIG-CSAM, many existing techniques from AI safety and secure & privacypreserving machine learning cannot be directly used for AIG-CSAM prevention, detection, and mitigation. Paradoxically, in part because child safety is an important and established area of concern, there are unique constraints and restrictions surrounding CS","cbCainxnrC1yl0pi","https://ap.wps.com/l/cbCainxnrC1yl0pi","pdf",2478663,4,1,21,"English","en",105,"# Introduction\n## Problem Scope and Motivation\n# Background\n## Generative AI Risks for Child Safety\n# Open Problems and Recommendations","[{\"question\":\"Why does preventing AI-facilitated child sexual abuse require new AI safety approaches?\",\"answer\":\"Existing AI safety methods assume data accessibility, transparency, and evaluation practices that conflict with the ethical and legal constraints around CSAM. These constraints create new technical challenges for auditing, red teaming, and fine-tuning-based prevention.\"},{\"question\":\"What risks does generative AI add for child safety in this work?\",\"answer\":\"Generative AI can be misused to create AI-generated child sexual abuse material (AIG-CSAM), facilitate child sexual exploitation, and reduce barriers to harm at unprecedented scale.\"},{\"question\":\"Which stages of the AI lifecycle are covered by the paper’s open problems and recommendations?\",\"answer\":\"The paper identifies 15 open problems across the AI development lifecycle, from dataset curation and model design to deployment and long-term maintenance, and provides recommendations for researchers, developers, and policymakers.\"}]",1784191647,53,{"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":86,"head_meta":88,"extra_data":90,"updated_unix":28},"position-preventing-ai-generated-csam-necessitates-new-approaches-to-ai-safety","",{"@graph":36,"@context":85},[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/position-preventing-ai-generated-csam-necessitates-new-approaches-to-ai-safety/83952/",{"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","2026-07-27","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why does preventing AI-facilitated child sexual abuse require new AI safety approaches?","Question",{"text":75,"@type":76},"Existing AI safety methods assume data accessibility, transparency, and evaluation practices that conflict with the ethical and legal constraints around CSAM. These constraints create new technical challenges for auditing, red teaming, and fine-tuning-based prevention.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What risks does generative AI add for child safety in this work?",{"text":80,"@type":76},"Generative AI can be misused to create AI-generated child sexual abuse material (AIG-CSAM), facilitate child sexual exploitation, and reduce barriers to harm at unprecedented scale.",{"name":82,"@type":73,"acceptedAnswer":83},"Which stages of the AI lifecycle are covered by the paper’s open problems and recommendations?",{"text":84,"@type":76},"The paper identifies 15 open problems across the AI development lifecycle, from dataset curation and model design to deployment and long-term maintenance, and provides recommendations for researchers, developers, and policymakers.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]