Beyond Infringement and Fair Use: Examining Generative AI, Unjust Enrichment and the Cultural Commons

The data-intensive architecture of Generative Artificial Intelligence (GenAI) relies on the unprecedented extraction of human expression from the open internet. The data gathered ranges from protected works to the public domain and cultural commons. Across jurisdictions, the courts are engaged in a binary debate of infringement or fair use. This paper argues that the binary outlook provided by the current copyright law overlooks the algorithmic enclosure of commons. The current frameworks allow for the aggregation of commons and the public domain by a privately controlled entity. To escape this doctrinal impasse, the paper evaluates AI training through the private law doctrine of unjust enrichment across comparative jurisdictions. The paper addresses the scale of appropriation of data by AI developers, the data that is scraped from the open web. Further, the paper explores the legality of such data usage and training of AI across the USA, UK and India. It then distinguishes public accessibility from the public domain and examines how cultural resources may be transformed into privately controlled computational capabilities. Applying the doctrine of unjust enrichment, the article analyses the applicability of the doctrine and the difficulty of valuation and quantification of enrichment. The three requirements of Unjust enrichment and its applicability, and the difficulty of fulfilling such requirements, are also explored.