CEPA has published a series examining how AI developers are harvesting copyrighted content from across the web — and offering suggestions of a path forward