Loops, Not Chains: Rethinking Value and Accountability in Generative-AI Creative Practice

Generative AI is reshaping cultural production by allowing images, sounds, styles, and stories to be deconstructed, recombined, and recirculated across media in looping, non-linear ways. Because AI-powered creativity is inherently collaborative and endlessly reworked, it is hard to decide who deserves credit—or who is liable—when harms such as offensive content or copyright conflict arise. While this openness broadens participation, it also poses new regulatory headaches. Most policymakers still picture AI development as a straight, step-by-step value chain that runs from data collection through model building to platforms and end-users (Botero Arcila, 2025). Creative practice, however, looks far messier and more circular. Using a social-constructionist lens, we treat generative AI not just as a tool but as a mediator, archive, and amplifier of meaning. Novelty now comes from remixing older material, documentary clips, selfies, logos, into fresh hybrids. With so many sources in play, creative ownership are unclear. Credit and responsibility no longer move in a line; they loop among creators, platforms, and audiences that keep remixing each piece. Although these loops enlarge creative participation, the financial upside still flows mainly to platforms and big brands (Edwards et al., 2024; Kenney & Zysman, 2018). We ask: (1) How do AI-mediated artefacts loop through creators, platforms, and audiences? (2) Who captures cultural and monetary value at each turn? (3) Where do accountability gaps surface when legal or ethical harms arise? To do so, we combine digital ethnography with computational trace analysis to follow the journey of three AI-mediated creative artefacts as they circulate online: the AI-mediated subversion of corporate imagery by the artist Str4nge Thing; the institutional co-optation of generative aesthetics by luxury brands, exemplified by Gucci’s Parallel Universes project; and Paolo Cirio’s use of AI to extract and politicize technical imagery in works like Capture, which reclaims surveillance footage of law enforcement. For each case, we trace how content is remixed, and recontextualized; we visualize the loops of circulation and who extracts value. Where possible, we complement this analysis with interviews and commentaries from artists, and remix communities. Our analysis will reveal recursive “value spirals” and power asymmetries. By mapping meaning, ownership, and (re)appropriation chains, we show how value extraction, responsibility, and agency shift in AI-driven culture, offering a framework for rethinking accountability beyond linear models.