AI developers pivot to licensing: Why Warner-Suno matters
Is the music industry re-thinking its approach to artificial intelligence? Against the backdrop of a settled copyright infringement claim, the recently reported deal struck between Warner Music Group (“Warner”) and AI song generator Suno – together with several similar licensing deals emerging across the creative sector – may signal a shift towards licensing of rights for AI training and outputs.
It has been reported that, under the arrangement with Warner, Suno can offer users the ability to generate music using the voices, names, likenesses, images, and compositions of Warner artists who choose to participate. Suno boasts “new creation experiences” and revenue streams, while Warner promises artists “full control” over the use of their identities and catalogues. Meanwhile, Suno will use licensed Warner content to train future models, and will introduce a paid-download system to formalise monetisation.
This move towards licensing by rights owners is not isolated. According to Semafor, Warner, Universal Music Group, and Sony Music Entertainment have all signed licensing agreements with the AI startup Klay Vision, granting access to their catalogues for model training. This makes Klay the first AI startup to publicly announce partnerships with all three major labels. Additionally, WMG and UMG have reportedly signed deals with Udio to collaborate on a new song creation service. Both of these arrangements were agreed as part of the settlement of copyright infringement litigation.
Some industry analysts see these agreements as more than settlements; they could set a precedent for future collaboration. Under this model, rather than contesting whether AI training and synthetic performance infringe copyright, publicity rights, or contracts, parties implement frameworks to streamline consent, compensation, and control through licences. Similar developments can be seen in other creative sectors, such as video, where archives are being licensed for generative model training. For rights owners, does this signal that the focus is shifting from litigation to longer term content licensing agreements?
A shift towards licensing has implications for all involved. For AI developers, different rights categories may begin to emerge for training purposes, such as voice likeness or compositional reuse. Depending on the terms of the licensing deals agreed with rights holders, these categories may have separate permissions and economics. Rights holders may be given more granular control mechanisms, including opt-in systems and revocation mechanisms, to enable them to control how AI providers use their content.
