UK music creators on AI: 91% want consent, 93% want payment, and the producers who use it most agree

UK Music's annual This Is Music 2025 report is mostly about money: the industry's contribution to the economy, exports and jobs. The 2025 edition also asked creators in its annual survey a set of questions about AI. On 26 Nov 2025 the Musicians' Union pulled out the headline numbers and tied them to its campaign for new law.

The headline numbers are lopsided. 91% of respondents say consent should be sought before compositions and recordings are used to train AI models. 93% say AI companies should pay when they do, and only 1% disagree. 92% want AI-generated music labelled so fans know what they are hearing, and 95% want their voice and image rights protected.

The more interesting part is in the tables split by role, and especially the producers.

Who answered, and when?

According to MusicRadar's report on the launch, UK Music surveyed 1,306 music creators, including songwriters, musicians, DJs and producers, between 20 March and 12 June 2025. The report breaks the AI answers into four groups: artists, performers, writers and producers. It doesn't print how many people answered in each group, so the role-level figures below are best read as direction, not precise measurement.

Are producers really using AI?

More than anyone else in the sample. Asked whether they currently use AI tools to help with music creation, 39% of producers said yes, they use or have used them. That compares with 20% of writers, 16% of artists, 9% of performers and 18% overall. Only 31% of producers said they had no intention of using AI, against 52% of all respondents.

The report says what that use looks like: producers and engineers already use AI for tasks such as audio restoration, demos and mastering. It also says producers have "a more positive view" of AI used where the human creator stays in control.

The attitude numbers back that up. When the question was about AI that supports the creative process while the human stays fully in control, 48% of producers felt positive and 31% negative. Across all respondents it was the reverse: 32% positive, 48% negative. 46% of producers agree that AI tools help their career, against 25% overall.

So do producers want looser rules on training?

No, and this is the finding that matters. The producers, the group most comfortable with AI, back consent and payment almost exactly as strongly as everyone else:

• Consent before training: 90% of producers agree, against 91% overall and 94% of writers.
• Payment for training: 92% of producers agree, against 93% overall.
• Labelling: 90% of producers agree, against 92% overall.
• Voice and image rights: 92% of producers agree, against 95% overall.

And when the question changes to AI that creates compositions, lyrics and recordings "with little to no human input", producers turn against it too: 66% negative and 15% positive. 61% of producers still say AI poses a threat to their career, not far behind the 66% of all respondents who say so.

In other words, the line these creators draw isn't for or against AI. It runs between tools they control and systems trained on their work without asking.

What are the unions and UK Music asking for?

The MU, which says it represents over 36,000 musicians, sets out five asks in its AI briefing: explicit consent before works are used for training, clear labelling of AI-generated works, a fair share of revenue from AI-generated music for composers and performers, a published and regularly updated record of the works AI firms trained on, and stronger publicity, personality and personal data rights that belong to the individual, so that AI companies, labels or other rightsholders can't exploit them without that person's consent.

UK Music's own list at the end of the report is close to that. It asks the government to reject any copyright exception that undermines how music is used and paid for, to require AI developers to disclose the copyright works they train on, to make labelling of AI music a legal requirement, and to "explore personality rights" covering image, likeness, voice and personal identity.

What has the government done since?

On 18 March 2026 the government published its Report on Copyright and Artificial Intelligence, required by section 136 of the Data (Use and Access) Act 2025. Its consultation drew 11,520 responses. The option it had originally preferred, a broad data mining exception with an opt-out, was supported by 3% of respondents, and the report says it is "no longer the government's preferred way forward". It quotes UK Music's response that consent "must be obtained, never assumed".

But the report doesn't change the law. On transparency it proposes to keep monitoring other countries' rules and to develop best practice with industry. On digital replicas it says existing protections don't cover every case where someone's likeness is copied without consent, and that it will consider whether a new digital replica or personality right would help. That is a promise to look, not a right that a session singer can rely on today. Compare Mexico, which put performers' consent for AI voice cloning into law in May 2026.

What we think

The survey quietly removes one of the AI industry's favourite arguments. The line goes that musicians already use AI, so they can't object to how it is built. UK Music's tables show the people using AI most are not giving up their rights in exchange. A producer who uses an AI mastering chain or a stem splitter hasn't agreed to have their catalogue scraped, and 90% of producers say so. Policymakers should treat "creators are adopting AI" and "creators consent to unlicensed training" as two separate claims, because the data says they are.

It also points to who loses most if nothing changes. Performers were the least likely to use AI (9%) and the most negative about it even when a human is in control (55%). Their income depends on their voice and their playing, which is exactly what a digital replica copies. The report also singles out songwriters and composers in media and production music as especially exposed. These are the people a personality right would protect first, and the government is still only exploring one.

The winners are the tools and companies that build on the split producers have drawn: assistive features, licensed training data, and the human making the decisions. Browser-based tools such as Veena, which uses AI to assist production while the artist stays in charge of the music, sit on the side of that line that producers already feel positive about. AI developers that rely on unlicensed training, by contrast, face a creator base that is unusually united: on payment, only 1% of respondents disagree.

Our advice for UK artists: keep records of what you release and where, so you can act if training disclosure ever becomes law. Read any contract clause that mentions voice, likeness or AI before you sign it, because for now the contract may be the only protection you have.