The future of AI music is here, but who gets paid when algorithms learn from our favorite songs? A newly released report, commissioned by the International Federation of the Phonographic Industry (IFPI), throws its weight behind a 'bilateral' licensing approach for AI-generated music. This means individual deals between rights holders and AI companies, a move that could significantly impact how artists and labels are compensated in this rapidly evolving landscape.
The report, titled ‘Generative AI Models at the Gate’ and conducted by the economic consulting firm Compass Lexecon, essentially argues that music rightsholders must be fairly compensated when their work is used to train these powerful AI models. It emphasizes the right of these rightsholders to refuse the use of their protected content for AI training purposes altogether. The report further dismisses arguments supporting free use, claiming them to be fundamentally flawed.
These arguments are not new; the music industry has been actively lobbying for fair compensation in the AI space. But here's where it gets controversial... the report places significant emphasis on 'bilateral licensing' – direct, one-on-one negotiations and agreements between individual rights holders (think record labels, publishers, and even individual artists) and AI music companies. We've already seen this in action with companies like Suno, Udio, Stability AI, and Klay Vision, who have been striking deals directly with rights holders in recent weeks.
And this is the part most people miss... While bilateral licensing seems to be gaining traction, there's another potential model on the table: collective licensing. This involves a collective management organization (CMO), like ASCAP or BMI, negotiating on behalf of many rights holders simultaneously. The IFPI report, however, is not a fan. It raises concerns about potential under-remuneration and anti-competitive practices that could arise from mandatory collective licensing. Bilateral deals, the report argues, offer "greater flexibility in setting terms and conditions.” In other words, rights holders can potentially negotiate better deals individually than they could through a blanket collective license.
But is this really the best approach? Could bilateral deals favor larger rights holders with more bargaining power, leaving smaller artists and independent labels at a disadvantage? The report highlights “obvious risks of under-remuneration and risks to competition that would arise from compulsory collective licensing”.
What do you think? Should individual artists and labels have the power to negotiate their own deals with AI companies, or is a collective licensing approach fairer and more efficient? Could a hybrid model, combining elements of both, be the ultimate solution? Share your thoughts in the comments below! You can read the full Compass Lexecon report here: (https://compass-lexecon.files.svdcdn.com/production/editorial/2025/04/Generative-AI-Models-at-the-Gate-Report-for-IFPI-Compass-Lexecon.pdf?dm=1743758320)
