ZOETROPE: A Decentralized Framework for Licensing Visual Content for Generative AI

Generative AI models are commonly trained on large-scale datasets of visual content scraped from diverse sources across the open internet. Yet, creators currently lack effective ways to express consent or to be recognized and rewarded when their work is included in such datasets. We present ZOETROPE, a decentralized framework that enables the discovery and licensing of visual content for AI training and usage. Viewing the copyright and AI challenge as a content supply chain problem, we show how open media provenance standards can be extended to support fair and transparent reuse of visual content in AI pipelines. ZOETROPE combines watermark-based provenance persistence, cascading licensing for downstream residual payments, and privacy-preserving visual search across decentralized repositories. By integrating discovery, creator consent, licensing, and compensation without recourse to a central marketplace, ZOETROPE offers a pathway toward more transparent and economically aligned relationships between creators and AI developers in the age of generative AI.