Music has always absorbed new technology — from the electric guitar to the synthesizer, the sampler and digital audio workstations. AI is the latest, and possibly the most disruptive, because it can generate entire songs from a text prompt. The industry is simultaneously adopting AI tools, licensing them, and suing some of their makers.

AI tools musicians already use

Much of the AI in music is uncontroversial and has quietly become standard:

  • Stem separation: splitting a mixed recording into vocals, drums, bass and other instruments. The technique made it possible to isolate John Lennon’s voice from an old demo tape and complete “Now and Then”, released by the Beatles in 2023 — the song later won a Grammy.
  • Mixing and mastering assistants that analyze tracks and suggest equalization, compression and loudness settings.
  • Audio repair: removing noise, clicks and room echo.
  • Composition aids: generating chord progressions, melodies or drum patterns as starting points.
  • Transcription: turning audio into sheet music or MIDI.

Song generators

Services such as Suno and Udio generate complete songs — instruments, vocals and lyrics — from short text descriptions. Quality improved rapidly, and millions of tracks have been created with them. They allow anyone to make music, but they also raise fundamental questions about where their training data came from and what happens to the market for human-made music.

In 2024, major record labels, coordinated by the Recording Industry Association of America, sued Suno and Udio for copyright infringement, alleging their models were trained on copyrighted recordings without permission. The disputes have since moved toward licensing: some labels reached settlements and partnership deals with AI music companies under which future models would be trained on licensed music with artist participation. The outcome is shaping how the business of AI music will work.

Separately, the US Copyright Office has said that purely AI-generated material is not protected by copyright, while human creative contributions to AI-assisted works can be.

Voice cloning and artists’ likeness

AI can imitate a singer’s voice convincingly. In 2023, a viral track imitating Drake and The Weeknd, created without their involvement, was removed from streaming services after objections. Lawmakers have responded: Tennessee’s ELVIS Act (2024) explicitly protects musicians’ voices against unauthorized AI imitation, and other jurisdictions are considering similar rules. Some artists, by contrast, have licensed AI versions of their voices and shared revenue with fans who use them.

Streaming platforms

Streaming services face floods of AI-generated tracks, sometimes uploaded in bulk to collect royalties or used with fake streams. Platforms have introduced policies against impersonation, spam uploads and artificial streaming, and some label or disclose AI-generated music.

What it means for musicians

AI lowers the cost of producing professional-sounding music and gives artists powerful new tools. It also increases competition for listeners’ attention and threatens income from production music for advertising, games and video, where “good enough” generated tracks may substitute for commissioned work. Live performance, artistic identity and direct relationships with fans become even more valuable.

What to watch next

  • Licensing frameworks between labels and AI music companies.
  • Laws protecting voice and likeness across more jurisdictions.
  • Streaming platform policies on AI tracks and royalty fraud.
  • Court decisions on training AI with copyrighted recordings.

Key takeaways

  • Stem separation, mixing and audio repair are widely accepted AI tools in music.
  • AI stem separation helped complete the Beatles’ “Now and Then” in 2023.
  • Song generators like Suno and Udio were sued by labels in 2024, with disputes moving toward licensing deals.
  • Tennessee’s ELVIS Act protects musicians’ voices against unauthorized AI imitation.
  • AI pressures production music income while raising the value of artistic identity and live performance.

Frequently asked questions

Can I release a song made with an AI music generator?

Check the service’s terms, which govern ownership and commercial use. Purely AI-generated material may not be protected by copyright in the US, and distributors may have their own rules.

Using someone’s voice without permission can violate publicity rights and laws such as Tennessee’s ELVIS Act, as well as platform policies. Rules vary by jurisdiction.

Will AI replace musicians?

AI can generate background and functional music cheaply, but audiences connect with artists as people. The bigger impact is on specific income streams such as production music.