Music Labels Push for Stronger AI Training Controls in New Negotiations
Ongoing talks with tech platforms highlight growing pressure to define how copyrighted material is used in generative systems
The music industry is moving to tighten its position on artificial intelligence, and the outcome could shape how other creative sectors approach the same problem. According to reporting in the Financial Times, major record labels including Universal Music Group, Sony Music, and Warner Music have been pushing for stronger controls over how their catalogs are used in training generative AI systems.
The negotiations are part of a broader effort to establish clearer boundaries around data usage. Labels are seeking agreements that would require technology companies to obtain licenses before incorporating copyrighted material into training datasets, as well as mechanisms to track and compensate for downstream use.
The issue has become more urgent as AI-generated music continues to gain traction online. Advances in voice synthesis and composition tools have made it possible to produce tracks that closely resemble the style or sound of well-known artists. While some of this content is experimental, other examples have reached large audiences, raising concerns about both revenue loss and brand dilution.
Labels are not rejecting the technology outright. Instead, they are attempting to define the terms under which it can be used. That approach mirrors earlier shifts in the industry, from the rise of digital downloads to streaming platforms. In each case, the initial disruption was followed by negotiations that established new economic frameworks.
The difference with AI is the role of training data. Unlike distribution models, which focus on how content is delivered, generative systems depend on large datasets that may include copyrighted material. Determining how that material can be used—and how rights holders are compensated—remains a complex legal question.
The outcome of these negotiations will have implications beyond music. Film, television, and publishing industries are facing similar challenges, particularly as generative tools expand into video and text. The frameworks established in one sector could influence how others approach licensing and enforcement.
For technology companies, the stakes are equally high. Access to high-quality data improves the performance of AI models, but obtaining that data through licensing agreements introduces cost and complexity. Balancing those factors will be critical to long-term viability.
The current discussions are unlikely to produce immediate resolution. Legal cases related to AI training are still working their way through courts, and regulatory approaches vary by jurisdiction. In the meantime, negotiations between labels and tech platforms are shaping the practical boundaries of what is acceptable.
The music industry has often been the first to confront changes in how content is created and distributed. Its response to AI may once again provide a template for how other parts of the entertainment business adapt.
For now, the message from labels is consistent. Access to creative work is not assumed. It is negotiated.