‘No children were harmed’: aesthetic infrastructures and ethical implications of AI-generated portraits

This article critically examines the Danish NGO Children’s Wellbeing’s 2023 campaign Stop Violence Against Children, which employed AI-generated portraits of children to visualise the effects of abuse. While the NGO framed the campaign as an ethical alternative to photographing real children – since ‘no children were harmed in the design process’ – we argue that using AI-generated images was not a neutral and innocent solution. By placing this campaign alongside the first Danish criminal case involving AI-generated child sexual abuse material, we highlight unsettling parallels between two different uses of photorealistic AI imagery. Drawing, among other things, on images’ operative and relational dimensions, our analysis shows how the AI-generated portraits actively engage with cultural, technical, and social infrastructures that renders it very likely that children were harmed in the extended design process. By unpacking the campaign’s visual rhetoric, specific technical creation, and cultural embedding, the article demonstrates that AI cannot serve as a simple ethical safeguard, but that the campaigns’ ethical intentions conflict with ‘The Civil Contract of Photography’ as elaborated by Ariella Azoulay. We conclude that the campaign's AI-generated images cannot be understood apart from the aesthetic structures and ethical traditions of photography, even if the NGO attempts to do so.