Dark Brand: The AI-Assembled Brand Image as a New Object of Brand Management

This paper introduces the concept of dark brand — the brand image that AI systems automatically assemble from publicly available internet content and deliver to consumers without brand-owner participation or review. As generative AI platforms increasingly serve as primary entry points for consumer purchase decisions, every brand now possesses a parallel identity constructed by large language models through statistical aggregation of web content. Consumers receive this AI-generated version daily through conversational interfaces, yet brand owners have neither created nor approved it. Drawing on brand equity theory (Aaker 1991; Keller 1993) and consideration set research (Wright & Barbour 1977), we develop a three-party cognitive split model describing the structural divergence among brand self-perception, consumer perception, and AI-mediated presentation. We decompose the dark brand into five observable dimensions and illustrate the framework through cross-platform data collection on BMW across four major Chinese AI platforms. Our analysis identifies a priority inversion: for brands in AI media, high visibility combined with low accuracy poses greater risk than low visibility alone. We formalize this through a visibility-accuracy quadrant and propose a measurement framework — the Brand AI Competitiveness Index — along with a two-phase management pathway. We conclude by discussing limitations and directions for empirical validation.