Answers are grounded in this paper and cite their sources.
Generative AI fashion design platforms are frequently promoted as expanding creative access and enabling more diverse forms of visual representation through personalized fashion imagery. In this article, we examine how AI fashion platforms negotiate recognizability, embodiment and norm-critical fashion when prompted to generate non-normative fashion imagery. Drawing on a qualitative study combining AI-augmented walkthroughs, practice-based design experimentation and reflexive analysis across four AI fashion platforms, we introduce a Glitch-Informed Representational Stress Test (GIRST) to examine the representational limits of these systems. The analysis demonstrates that AI fashion platforms consistently stabilize normative aesthetic conventions and forms of embodiment, while non-normative representations frequently appear through inconsistencies, partial disruptions or unstable outputs that remain difficult to sustain across iterations. These inconsistencies and unstable outputs can be understood as glitches that expose the limits of what AI fashion platforms can render as recognizable. We conceptualize the corrective work required to negotiate these limits as glitch labour: the iterative labour through which users attempt to sustain non-normative representation within AI fashion platforms. The article contributes to critical fashion and platform studies by demonstrating how AI fashion platforms do not simply constrain non-normative representation, but actively shape the conditions under which embodiment becomes recognizable within algorithmically mediated design environments.
No takes yet. Share an insight, caveat, or question.
Christoforidou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: