'Everybody Is On It and Nobody Wants to Talk About It': AI's Secret Takeover of the Music Industry

The music industry has been running a secret experiment. And the results are in.

A comprehensive Rolling Stone investigation, surfaced by The Decoder, has documented what many inside the industry already knew but wouldn't say publicly: top producers, engineers, and songwriters are extensively integrating AI music generation tools into their creative processes — and maintaining deliberate silence about it. The characterisation from Suno CEO Mikey Shulman has circulated widely since the report: AI is "the Ozempic of the music industry — everybody is on it and nobody wants to talk about it."

The Scope of Secret Adoption

Sonarworks research found that seven in ten music professionals — producers, engineers, and songwriters — experiment with AI tools at least occasionally. One in five uses them regularly. The applications range from established AI-adjacent tasks like audio restoration and stem separation to more fundamental creative work: AI-generated backing tracks, demo production, and — most significantly — AI-generated samples used in lieu of licensed recordings or hired studio musicians.

The shift is most dramatic in hip-hop. Jay-Z's longtime sound engineer Young Guru estimates that more than half of sample-based hip-hop production now relies on AI-generated retro samples rather than original licensed recordings or live studio musicians. This is not a fringe phenomenon. It represents a fundamental transformation in how a core genre of contemporary music is produced.

Why the Silence

The precedent that explains the silence is Teddy Swims. The singer faced significant public backlash after openly admitting AI use in his creative process — a cautionary tale that circulated through the industry and established the informal norm: use AI, don't say so. The incentive structure is clear. There is no upside to disclosure, and there is visible downside.

For established songwriters, the economics of AI adoption are genuinely attractive. Songwriter Michelle Lewis describes the technology as "empowering" in specific structural terms: it eliminates copyright splits (AI-generated music has no co-writers to negotiate with), reduces production costs, and compresses demo creation from hours to minutes. One major artist reportedly recorded a demo generated within moments from a lyrics-and-chords input. The creative speed advantages are real.

Who Bears the Cost

The efficiency gains for established professionals come directly at the expense of the lower tier of the music economy. Session musicians are seeing fewer gigs. Studio assistants have less work. The production music and stock music markets — which supply soundtrack material for smaller television productions, podcasts, and digital content — have, by multiple accounts, become "practically toast." Entry-level positions in the music industry are disappearing faster than the industry is acknowledging.

The talent pipeline concern mirrors the engineering career ladder problem surfaced in software development: the entry-level work that trains the next generation of skilled professionals is being automated away before that generation can learn from it. Music production, like software engineering, depends on tacit knowledge accumulated through doing lower-level work. When that work disappears, so does the formation of the people who would have done it at scale.

The Legal Vacuum

The adoption is outpacing the legal framework by years. Copyright protection for AI-generated content remains legally unsettled. Detection software capable of identifying AI-generated music does not reliably exist — major labels, by multiple accounts, rely on an informal honour system. A Suno investor recently admitted publicly that the music generator directly competes with human musicians, a statement that may prove legally significant in ongoing copyright litigation but has not slowed deployment.

Google's Lyria 3, released last week, has positioned its training data approach around licensed and permission-based music — a deliberate differentiation from Suno and Udio, which are both facing legal challenges from major labels. Whether permission-based training represents a legally durable position will depend on courts that have not yet definitively ruled on the underlying question.

The Asymmetric Challenge

The final dynamic is the hardest to resolve. Human musicians and songwriters are competing against AI-using peers who operate largely invisibly, face no comparable cost structure, and have no incentive to disclose. The competitive disadvantage is structural: there is no transparency mechanism, no labelling requirement, and no clear regulatory framework that would change the incentive calculus in the near term.

The Ozempic comparison is apt in ways beyond the intended irony. Ozempic reshaped the weight loss industry invisibly, produced winners and losers that were structurally determined rather than individually chosen, and created a regulatory and social adjustment that is still ongoing. AI music generation is operating on the same pattern — adopted broadly, discussed narrowly, and reshaping an entire creative economy before the affected parties have established the institutional language to describe what's happening to them.