Will AI replace musicians? Which music jobs are really at risk

The question usually gets two lazy answers. One says AI can't feel, so musicians are safe. The other says AI already sounds human, so musicians are done. Both skip the part that decides careers: who is paying, and what exactly they are paying for. When a buyer wants a person, AI struggles. When a buyer wants 90 seconds of upbeat music for a video by Friday, AI is already good enough.

How much money is at risk for musicians?

The most cited estimate comes from a December 2024 study by PMP Strategy for CISAC, the global body of authors' societies. It models what happens to songwriters' and composers' income as AI-made music spreads, compared with a world without generative AI. It is a projection, not a measured loss, and it covers songwriters' royalties, not recording artists.

• 24%of music creators' revenues at risk by 2028 (CISAC/PMP Strategy, Dec 2024)CISAC
• €10bnprojected cumulative loss for music creators, 2023 to 2028CISAC
• 60%of B2B library music expected to be AI-generated by 2028Music Business Worldwide (CISAC study)
• €950min songwriter royalties at risk in Germany and France in 2028 (Goldmedia for GEMA and SACEM, Jan 2024)GEMA

Which music jobs are most at risk from AI?

Exposure depends on one question: does the buyer care who made it? Here is our assessment, grounded in the CISAC projections and in how AI music is actually being bought and streamed in 2026.

Why are some musicians safer than others?

Because music is two products sold under one name. One is sound: something to fill a video, a store or a study session. The other is a person: a voice you know, a story you follow, a show you travel to. AI is very good at the first product and has not cracked the second. Even the AI acts that charted were built around human creators: poet Telisha Jones writes the lyrics for Xania Monet, according to CBS News.

The law also leans toward people. The U.S. Copyright Office said in January 2025 that prompts alone do not make an output copyrightable, while human authorship in AI-assisted work still counts. A label that wants to own a catalog has a strong reason to keep humans in the loop.

How does AI music cut into streaming income?

Most streaming services pay rights holders from a shared pot, split by each track's share of total streams. Every stream that goes to an AI track is a slice that does not go to a human one. That is why fraud matters so much: Deezer says up to 85% of streams on fully AI tracks in 2025 were fake, generated by bots to farm royalties from the same pot human artists draw on.

Platforms are starting to wall that off. Deezer strips fraudulent streams out of royalties and keeps fully AI tracks out of its recommendations. Tidal said in June 2026 it would stop paying royalties on fully AI-generated music. Spotify, which only pays tracks that pass 1,000 streams a year, said in October 2026 that tracks likely made with AI are 10 times more likely to fall below that line. In the Ipsos survey, 69% of respondents said fully AI music should pay less than human music. If that view spreads, the money argument for flooding services with AI tracks gets much weaker.

What are musicians doing about AI?

• Signing statements. The Statement on AI Training, which calls unlicensed training an unjust threat, had 50,544 signatures in October 2026, including Thom Yorke, Kate Bush and Björn Ulvaeus.
• Protesting loudly, and quietly. In February 2025, 1,000 UK musicians released Is This What We Want?, an album of near-silent empty studios, against a UK plan to let AI firms train on their work.
• Changing the law. Tennessee's ELVIS Act, signed in March 2024, added a person's voice to the rights protected from unauthorized AI copies.
• Suing and licensing. Labels sued Suno and Udio in 2024; Warner then licensed Suno with an opt-in for its artists' likeness and songs, while Universal and Sony kept suing.
• Using the tools. Luminate found 18% of US musicians already use AI to edit or remix music (2026).

What should a working musician do now?

Move toward the parts of the job that are about you. Build a direct line to listeners, play live, put your name and face on your work, and register your songs so you are paid when licensing money starts flowing. Use AI where it saves you time, the way producers adopted drum machines and Auto-Tune, but don't compete with it on price for anonymous background music. That race has a winner already.

For the bigger picture, read is AI music a fad?, our breakdown of AI music vs. human music, and how royalties work in AI music royalties for creators.