Unjust enrichment could become AI's data remedy, say researchers

A new academic paper from researchers Yangzi Li and Jyh-An Lee, set to appear in the Common Law World Review (Vol. 55, 2026), proposes unjust enrichment as a more effective legal framework for addressing generative AI's unauthorized use of protected data. Current intellectual property (IP) and privacy laws struggle to handle the scale and opacity of training data ingestion, leaving data owners with limited recourse. The authors argue that unjust enrichment — a doctrine focused on recovering benefits obtained unfairly — avoids the burdens of proving infringement or privacy violations, which often fail in AI cases due to the transformative nature of model training.

The paper highlights that gain-based restitution under unjust enrichment offers distinct advantages: it shifts the legal question from 'was the data use wrongful?' to 'did the AI developer obtain a measurable benefit from the data?' This approach could lower litigation barriers for data owners while providing clearer liability rules for AI companies. By sidestepping difficult debates about data ownership and fair use, unjust enrichment may offer a more pragmatic solution that protects data creators without stifling AI innovation. The research adds to a growing body of work on legal remedies for the AI training data crisis.