UNICEF - Artificial Intelligence and Child Sexual Abuse and Exploitation

Artificial Intelligence and Child Sexual Abuse and Exploitation This brief explores how generative AI is creating new threats to children’s right to protection from sexual abuse and exploitation. It examines emerging risks and evidence and outlines urgent priorities for action.

ISSUE BRIEF - FEBRUARY 2026

2. Urgent action is needed by governments and industry to prevent the creation and spread of AI-generated sexual content of children.

3. Parents, educators, social services, mental health professionals, and law enforcement need resources and continuous training to support impacted children.

KEY MESSAGES

1. The growing prevalence of AI-powered image or video generation tools that produce child sexual abuse material marks a significant escalation in risks to children through digital technologies.

The challenge The rise of powerful AI image generation tools has transformed the risk landscape for child protection. Less than 5 years ago, high-quality generative models were largely proprietary and required significant computing power and expertise.1 Today, opensource models can run on consumer-grade hardware, making it far easier for perpetrators to create sexual abuse content.22 Even when no ‘real’ child is directly involved, the individual and societal harm is real – such content normalises the sexualisation of children, can fuel demand, and victimises children whose likeness is used.3 It also complicates victim identification by making it harder to determine whether an image depicts a real child in urgent need of help, potentially wasting already scarce law enforcement resources and delaying responses to safeguard children.4 These developments mark a profound escalation of the risks children face in the digital environment. Perpetrators can create realistic sexual images of a child without their involvement or awareness, meaning that a child can now have their right to protection violated without ever sending a message or even knowing it has happened. Children then face the full weight of the consequences: shame, stigma, moral judgement from peers and adults, social isolation, and long-term emotional harm. What do we know? Increased accessibility of AI-powered image or video generation tools has led to

violated without ever sending a message or even knowing it has happened. Children then face the full weight of the consequences: shame, stigma, moral judgement from peers and adults, social isolation, and long-term emotional harm. What do we know? Increased accessibility of AI-powered image or video generation tools has led to an increase in the production and spread of child sexual abuse materials (CSAM). These risks are not hypothetical; they are documented and growing at a rapid pace. For instance, in one month, the UK’s Internet Watch Foundation (IWF) found nearly 14,000 suspected AI-generated images on a single dark-web forum dedicated to child sexual abuse materials, almost a third confirmed as criminal, and the first realistic AI videos of child sexual abuse.5 IWF also identified AI CSAM on mainstream platforms. Beyond content created by adult offenders, AI generated CSAM also includes ‘deepfake’ nudes created in peer-to-peer contexts, often disproportionately targeting girls. In Korea, for instance, law enforcement reported a ten-fold increase in sexual offenses involving AI and deepfake technologies between 2022 to 2024, with teenagers constituting the majority of the accused.6 In the US, Thorn’s survey found that 1 in 10 teens knew of cases where friends or classmates had created synthetic non-consensual intimate images of other children using generative AI tools.73 Recent large-scale research by UNICEF, ECPAT and INTERPOL under the Disrupting Harm project8 showed that across 11 countries, at least 1.2 million children reported having had their images manipulated into sexually explicit deepfakes through AI tools in the past year. In some of the countries, this amounts to 1 in 25 children, or the equivalent of one child in a typical classroom. Children themselves are already deeply aware of this threat: in some of the 11 countries, up to two thirds said they worry that AI could be used to create fake sexual images, though levels of worry vary hugely from one country to another, underscoring the need for contextually adapted awareness and protection measures.

Children themselves are already deeply aware of this threat: in some of the 11 countries, up to two thirds said they worry that AI could be used to create fake sexual images, though levels of worry vary hugely from one country to another, underscoring the need for contextually adapted awareness and protection measures. Increasingly, the use of AI-powered image and video generation tools go beyond just creating CSAM. The US-based National Center for Missing and Exploited Children (NCMEC) report how generative AI is linked to various forms of child sexual exploitation, including grooming and extortion.9 In summary, these technological developments mean that a child’s body, identity, and reputation can be violated remotely, invisibly, and permanently. Furthermore, children can be threatened, blackmailed, or extorted using images that are fabricated, but that nevertheless feel real, carry social credibility, and can follow them for decades. Key priorities for action This unprecedented situation poses new challenges for prevention and education, legal frameworks, and response and support services. Current prevention efforts, which often focus on teaching children about online safety and the risks of creating or sharing sexual images, are still important but insufficient when sexual content can be artificially generated. • Parents and caregivers need to be informed about AI-enabled sexual exploitation and abuse and have the understanding and capacity to support children who are affected. • Schools, as frontline environments, should educate students about AI-related risks such as deepfake nudes and ‘nudify’/’undress’ AI tools, including the significant harm it causes to those affected, encourage reporting of harmful behaviour, and train educators to respond appropriately.104 • Social services, mental health professionals, and law enforcement need resources and continuous training to support impacted children, recognising the unique psychological impacts of AI-generated child sexual abuse materials, including when produced by peers.11 • Crucially, the threat of AI-generated sexual content requires urgent legislative action. National criminal laws must be updated and enforced for the digital age.12 This includes ensuring definitions of CSAM cover AI-generated material, aligning

chological impacts of AI-generated child sexual abuse materials, including when produced by peers.11 • Crucially, the threat of AI-generated sexual content requires urgent legislative action. National criminal laws must be updated and enforced for the digital age.12 This includes ensuring definitions of CSAM cover AI-generated material, aligning national frameworks with international standards, and guaranteeing remedies for child victims. States should further ensure the availability of a range of criminal, civil and administrative sanctions for legal persons for offences relating to child sexual exploitation and violations of obligations to protect children from such harms.13 • Finally, increased industry action, accountability and transparency are essential. States should require companies to conduct child rights due diligence, particularly child rights impact assessments.14 Companies should incorporate robust disclosure on child rights impacts in their regular reporting.15 Every actor in the AI value chain, from dataset providers to model developers, must embed safety-by-design. This includes pre-release safety testing for open-source models to reduce misuse or illegal use,16 with emerging standards offering guidance.17 Industry also has an immeasurable role to play in contributing to the development of solutions to support child rights and well-being in the age of AI, including by contributing to cross-industry initiatives and research efforts.5 Endnotes 1 Caoilte Ó Ciardha et al., ‘AI Generated Child Sexual Abuse Material -- What’s the Harm?,’ arXiv:2510.02978, preprint, arXiv, October 3, 2025. 2 Ibid. 3 Ibid.; UN Committee on the Rights of the Child, Guidelines regarding the implementation of the Optional Protocol to the Convention on the Rights of the Child on the sale of children, child prostitution and child pornography, CRC/C/156, 10 September 2019, para. 63. 4 Caoilte Ó Ciardha et al., p. 17. 5 IWF, ‘2024 Update: Understanding the Rapid Evolution of AI-Generated Child Abuse Imagery,’ 2024; IWF,

para. 63. 4 Caoilte Ó Ciardha et al., p. 17. 5 IWF, ‘2024 Update: Understanding the Rapid Evolution of AI-Generated Child Abuse Imagery,’ 2024; IWF, ‘What has changed in the AI CSAM landscape?,’ July 2024; IWF, ‘AI becoming ‘child sexual abuse machine’ adding to ‘dangerous’ record levels of online abuse, IWF warns,’ 16 January 2026. 6 ‘Deepfake, other digital sex crimes by teens nearly double in South Korea in four years,’ The Straits Times, 15 September 2025. 7 Thorn, ‘REPORT: 1 in 10 Minors Say Peers Have Used AI to Generate Nudes of Other Kids,’ 14 August 2024. 8 The range of prevalence across the 11 countries was 0.4%-4.3%. Population-level estimates were calculated by using post-stratification weights, which calibrate the sample to UN 2024 age population totals, adjusted by survey-estimated child internet-use rates. Disrupting Harm is a multi-country research project about technology-facilitated child sexual abuse, implemented in 25 countries by UNICEF, ECPAT International and INTERPOL between 20192026, with funding from Safe Online. In 2025, approximately 11,000 children living in 11 countries were asked questions about AI-generated child sexual abuse through a nationally representative household survey. More data and findings from Disrupting Harm are forthcoming in 2026 and will be published here: https://safeonline.global/ disrupting-harm-2/ 9 NCMEC, ‘Spike in online crimes against children a “wake-up call”,’ 4 September 2025. 10 See, for example, eSafety Commissioner (Australia), ‘Deepfake damage in schools: How AI-generated abuse is disrupting students, families and school

9 NCMEC, ‘Spike in online crimes against children a “wake-up call”,’ 4 September 2025. 10 See, for example, eSafety Commissioner (Australia), ‘Deepfake damage in schools: How AI-generated abuse is disrupting students, families and school communities’, 27 June 2025. See also: Grossman, S., Pfefferkorn, R., & Liu, S, ‘AI-Generated Child Sexual

Abuse Material: Insights from Educators, Platforms,

Law Enforcement, Legislators, and Victims’, Version 1, Stanford Digital Repository, 2025. 11 See, for example, Internet Watch Foundation and CEOP Education (UK), ‘Child sexual abuse material generated by artificial intelligence: An essential guide for professionals who work with children and young people’, Internet Watch Foundation, 2025. 12 United Nations Children’s Fund (2022) ‘Legislating for the digital age: Global guide on improving legislative frameworks to protect children from online sexual exploitation and abuse’ UNICEF, New York (updated version to be released in late 2026). 13 Ibid., pp. 106-107. 14 UN Committee on the Rights of the Child, General comment No. 25 (2021) on children’s rights in relation to the digital environment, CRC/C/GC/25, 2 March 2021, para. 38. UNICEF has developed a toolbox for businesses with step-by-step guidance on conducting child rights impact assessments in relation to the digital environment. See UNICEF, D-CRIA Toolbox, 2025. 15 UNICEF, ‘Corporate reporting on child rights in relation to the digital environment: Disclosure recommendations and guidance for business’, 2025. 16 See further: UNICEF Innocenti – Global Office of Research and Foresight, UNICEF Guidance on AI and Children 3.0, UNICEF Innocenti, Florence, December 2025, p. 18.

mendations and guidance for business’, 2025. 16 See further: UNICEF Innocenti – Global Office of Research and Foresight, UNICEF Guidance on AI and Children 3.0, UNICEF Innocenti, Florence, December 2025, p. 18. 17 Thorn, ‘Safety by Design for Generative AI: Preventing Child Sexual Abuse,’ 2024. Acknowledgements This brief was drafted by Afrooz Kaviani Johnson, UNICEF Global Programme Division, and Daniel Kardefelt Winther and Anisha Bhargava, UNICEF Office of Strategy and Evidence – Innocenti. Thanks are extended to Josianne Galea Baron, UNICEF Global Programme Division for inputs on the draft.

Disclaimer: This is a working document. It has been prepared to facilitate the exchange of knowledge and to stimulate discussion. The text has not been edited to official publication standards and UNICEF accepts no responsibility for errors. The statements in this publication are the views of the author(s) and do not necessarily reflect the policies or the views of UNICEF. The designations in this publication do not imply an opinion on legal status of any country or territory, or of its authorities, or the delimitation of frontiers.Published by UNICEF Child Protection and Migration Practice, Global Programme Division and UNICEF Office of Strategy and Evidence – Innocenti Email: childprotection@unicef.org | innocenti@unicef.org Web: unicef.org | unicef.org/protection | unicef.org/innocenti | data.unicef.org

Social media: @UNICEF | @UNICEFInnocenti on Bluesky, Instagram, LinkedIn, and YouTube | UNICEF-Child Protection on LinkedIn Suggested citation United Nations Children’s Fund, Artificial Intelligence and Child Sexual Abuse and Exploitation, UNICEF, February 2026.

© United Nations Children’s Fund (UNICEF), [February] [2026]. Except for the UNICEF logo, photos and where another © appears, this work is licensed under CC BY 4.0. The UNICEF logo and enclosed photos may only be used in exact copies of the publication. Where another © appears, you must ask the listed copyright holder to use the item outside exact copies of the publication. About us UNICEF, the United Nations agency for children, works to protect the rights of every child, everywhere, especially the most disadvantaged children and in the toughest places to reach. Across more than 190 countries and territories, we do whatever it takes to help children survive, thrive, and fulfil their potential. UNICEF Office of Strategy and Evidence – Innocenti accelerates progress for children by working to ensure that policies and programming are informed by high-quality evidence. As the global custodian of child-related official statistics, it works closely with governments and partners to strengthen nationstatistical systems. Through world-leading data, research and foresight, it underpins UNICEF’s global leadership on children’s rights and serves as the organization’s hub for setting strategy and monitoring programmes. With the active engagement of young people and other partners, it supports advocacy and dialogue aimed at improving the lives of children everywhere.