USA – NEW YORK Trends and Developments Contributed by: Nancy E Wolff, Scott J Sholder and Elizabeth Safran, Cowan DeBaets Abrahams & Sheppard LLP
offering content licensing solutions include Protégé and Created by Humans. From collective licensing to individual licence arrange - ments between industry leaders and AI developers, it is clear that AI licensing will be a major part of the AI/ content ecosystem. As the marketplace rises to meet the moment, AI developers and start-ups have less recourse to claim that licensing content for training is impossible due to prohibitive cost and the large-scale amount of material required. Ultimately, whether driv - en by litigation or market considerations, the players within this space are increasingly recognising the pos - sibility of fostering AI technology in a way that fairly compensates rights-holders and accommodates their legal rights. Recommendations for best practices in AI-licensing Considering the many sectors increasingly affected by AI technology, AI licensing is far from a one-size- fits-all approach; however, various trends and best practices are emerging across industries. Fundamentally, rights-holders’ interests in maintaining an intentional say in how, and whether, their works are used in AI model training and development drives any consideration of best practices for licensing agree - ments. To this end, many licensing agreements estab - lish an opt-in approach for rights-holders – rather than the opt-out approach more likely to be preferred by AI developers – whereby rights-holders must affirma - tively agree to include their work in a training dataset, rather than proactively opt out of inclusion. It is also important to ensure that licensing revenue, whether lump sum, or otherwise, will be appropriately divided and shared with each group of rights-holders. Ensur -
ing transparency as to the copyrighted content con - tained within a dataset and crediting contributors thereof is also paramount. Considering that many works, such as musical or visual works, are globally administered through international agreements and licensing networks, consistency among AI licensing provisions, and accounting for international laws is also important to avoid disruption. Finally, rights-hold - ers should ensure that any agreement protects them from the unauthorised use of their name, image, like - ness, voice and performance via digital replicas and imposes guardrails against infringing output, where While 2025 was clearly another fast-paced year for generative AI development, it was also an encourag - ing year for content owners with the emergence of strong AI licensing standards, practices and deals between major media companies and AI developers. Throughout 2026, expect the market for AI licensing to continue to expand and strengthen. The slate of ongo - ing litigation surrounding the use of pirated content for AI training only throws the need for this market into starker relief, with settling cases establishing a prec - edent for collaboration and licensing between par - ties formerly on opposite sides of the “v”. Expect any determinations as to fair use to inform the progress of AI licensing; however, considering the increasing num - ber of key industry players and AI developers enter - ing into voluntary licensing agreements outside of or notwithstanding litigation, it certainly appears that AI licences are quickly becoming an entrenched cost of doing business for AI model developers, and that licensing will continue to play a crucial role in the con - tinued advancement of generative AI technologies. applicable. Conclusion
699 CHAMBERS.COM
Powered by FlippingBook