Record labels are licensing music catalogs to AI companies that they once accused of copyright infringement, turning a legal battle into a new commercial partnership. But musicians say the agreements may put their recordings, performances and creative identities to work without their informed consent. The unsettled question: who benefits when AI can make new music from existing work?
Key takeaways
The deals mark a shift from lawsuits to licensing, but their terms and the impact on individual artists remain unclear.
- Suno has launched models built with licensed catalogs from Warner Music Group, BMG and Believe; Udio has deals with Universal and Warner.
- Labels and platforms say participating artists can opt in and receive payment, but have not disclosed key details about rates or participation.
- Musicians and their union argue performers may be excluded from decisions and compensation.
From copyright lawsuits to licensing deals
In 2024, Universal, Sony and Warner sued AI music companies Suno and Udio, alleging their systems had been trained on copyrighted recordings. Settlements and partnerships followed: Warner signed with both platforms, Universal partnered with Udio, and further agreements brought in independent music companies. Suno’s latest models, released earlier this month, use licensed music from Warner, BMG and Believe.
The products differ. Udio’s model is designed as a closed environment where generated songs cannot be downloaded or distributed elsewhere. Suno allows downloads, with limits and watermarks, and says tracks can be distributed through Believe and TuneCore. Spotify has also agreed to deals with Universal and Merlin for a planned AI product.
Consent and payment remain unresolved
Companies have described opt-in arrangements for artists, with credit and payment, but have not revealed how much artists will earn, how many have joined or how much say each musician has. Rights are also divided: labels often own sound recordings, while songwriters and publishers may control compositions. Contract terms determine who can authorize what.
Warner’s deal with Suno allows participating artists’ names, likenesses, voices and compositions to be used in AI-generated music. Yet Suno’s product chief has acknowledged that label contracts shape what artists can do. It is also not clear in every case whether opting out excludes music from model training, from consumer-facing features, or both.
Musicians push back
The American Federation of Musicians sued Universal and Warner, alleging they licensed recordings featuring union members to Suno and Udio without informing or paying those performers. The labels have sought to dismiss the case and dispute the union’s claims. Separately, a proposed class action by Jason Isbell and other musicians argues that AI-generated songs can evoke performers’ identities without permission; Suno disputes those allegations.
For session musicians, the concern is practical as well as legal: a performance could be reused without a new negotiation or payment. Songwriters also warn that similarities in musical ideas can be harder to prove than the imitation of a recognizable voice or recording.
A new revenue stream, with old questions
Labels say licensing can create income for artists and songwriters. Supporters see it as a way for the music industry to engage with AI rather than leave rights unresolved. Critics fear the technology could weaken demand for working musicians, echoing longstanding arguments over how streaming revenue is shared.
For Power AI Media readers following AI’s impact on culture and work, the central issue is not simply whether AI music will grow. It is whether artists have meaningful choices—and transparent terms—when their work helps make these products valuable. With payouts still undisclosed and legal disputes ongoing, that test has yet to come.