The music business is once again being forced to confront a familiar kind of disruption, and this time the fault line runs through generative AI. Monica Corton, a veteran publisher, argues that the industry’s anxiety echoes the early days of Napster: new technology arrived quickly, rights holders were slow to adapt, and the commercial framework was built only after the damage was done. Her central warning is that AI companies have already created valuable products from songwriters’ work, while many independent publishers still see no direct compensation.

That frustration has sharpened as AI music companies begin to strike licensing deals. According to Music Business Worldwide, Musixmatch has agreed licences with Sony Music Publishing, Universal Music Publishing Group and Warner Chappell Music, giving it access to more than 15 million works for analytical and non-generative AI services. In a separate development, Klay Vision has secured agreements with Warner Music Group, Universal Music Group and Sony Music Entertainment, positioning itself as a licensed AI music startup trained solely on authorised material.

Corton says the real issue is not simply whether AI firms should pay for training data, but how royalties can be traced and divided once a model has been built. She argues that synthetic data only complicates the picture further, because it makes it harder to identify which songs shaped a given output. On her reading, every generated track is ultimately rooted in human-made music, which means the sector needs a system that can map influence back to source works.

That is where attribution technology becomes central. Music Business Worldwide reported that companies such as Musical AI, formerly known as Somms.ai, are building rights-management tools designed to identify how much of an output derives from which songs, and to help automate compensation. Corton says such systems will only be useful if publishers can audit them, since many compositions are owned by multiple writers and publishers in uneven shares. She also warns that the mathematics become far more complicated when one AI-generated track may draw on dozens of works at once.

Corton’s preferred model is one in which outputs stay inside a controlled environment rather than being scattered across streaming platforms. She points to UMG and WMG’s arrangement with Udio as an example of the kind of “walled garden” structure she believes is needed, arguing that leaving AI-generated tracks to circulate freely on services such as Spotify or Apple Music would distort royalty pools and make cross-platform accounting unworkable. Her broader conclusion is that the industry must learn how the technology works, use that knowledge to negotiate stronger terms, and avoid repeating the mistakes made during the digital transition.

Source Reference Map

Inspired by headline at: [1]

Sources by paragraph:

Source: Noah Wire Services