This article introduces the use of AI-replicas as an alternative to traditional anonymisation methods in image-based qualitative research. It emphasises the ethical and practical dilemmas posed by current anon-ymisation methods, such as distortion or loss of emotional and contextual information in images, and proposes the use of AI-replicas to preserve the integrity and authenticity of visual data while ensuring participant anonymity. The article outlines the technological foundations of generative artificial intelligence (AI) and the practical application of Stable Diffusion to generate AI-replicas for anonymisation and fiction-alisation purposes. Furthermore, it discusses the potential biases present in generative AI to suggest ways to mitigate these biases through careful prompt engineering and participatory approaches. The introduced approach aims to enhance ethical practices in visual research by providing a method that ensures participant anonymity without compromising the data's qualitative richness and interpretative validity. Ethical and research-practical challenges in using images for qualitative research Over the last three decades, there has been a growing interest in the visual dimensions of social life, and research methods that utilise visual data of some kind have become legitimate and