So you’ve discovered some shiny new AI music generator and you’re wondering whether to ditch old-fashioned human composers like me for good. Fair enough – those AI tools are genuinely impressive, and I’d be lying if I said I hadn’t had a play myself. But before you hit ‘generate’ and plaster that track across your content, there’s something you need to know.
And as a composer with a Master’s in Electronic Composition, a decade dealing with the minefield that is music copyright, and over 20 years in this business, I’ve got a front-row seat to a legal car crash that’s about to happen.
Remember: I'm not a lawyer, and this isn't legal advice. I'm a composer who wants content creators to make informed choices.
Let’s start with what AI music generators like Suno and Udio actually offer. They’re remarkable pieces of technology – type in a prompt, wait a few seconds, and out comes a fully produced track. It’s genuinely impressive, and I say that as someone whose entire career depends on the old-fashioned method of, you know, actually composing music.
The promise is seductive: unlimited music, instant generation, seemingly free (or cheap), no need to credit anyone. For content creators drowning in deadlines, it looks like the perfect solution.
But here’s the catch – and it’s rather a large one. The legal protections underlying AI-generated music are built on sand. Don’t take my word for it; take Suno’s own admission from their terms of service: “Due to the nature of machine learning, Suno makes no representation or warranty to you that any copyright will vest in any Output.” [1]
Read that again slowly. The company itself admits it can’t promise you’ll own the copyright to what you’ve generated.
It’s a bit like buying a house where the estate agent cheerfully admits they’re not entirely sure anyone actually owns the land. You might be fine. Or you might wake up one morning to find yourself in a legal quagmire. Rather a gamble, that.
I promise to make this as painless as possible, but you really do need to understand the legal landscape here. Skip this section at your peril – this is where it gets properly concerning.
In January 2025, the US Copyright Office released its definitive guidance on AI-generated content, and the news isn’t good for AI music generators. [2] The ruling is crystal clear: 100% AI-generated content cannot be copyrighted and falls into the public domain.
The Copyright Office stated: “The outputs of generative AI can be protected by copyright only where a human author has determined sufficient expressive elements.” [2]
Now, you might think, “But I wrote the prompt! That’s creative input!” Sorry, but no. Writing a prompt – even a detailed, clever one – does not constitute authorship according to copyright law. [3] This was definitively established in the Thaler v. Perlmutter case, which concluded that copyright protection is reserved for works of human creation. [4]
Let me translate what this means in practical terms: if you generate music using an AI tool like Suno, you cannot copyright that music. Full stop. Anyone can use it, copy it, or claim it as their own, and you have absolutely no legal recourse.
Here in the UK, things work slightly differently. Section 9(3) of the Copyright, Designs and Patents Act 1988 does provide some protection for “computer-generated works”, with copyright typically assigned to whoever made the arrangements for the work’s creation. [5]
For years the UK Intellectual Property Office was consulting on whether to scrap that, alongside a much bigger question: whether AI companies should be allowed to train on copyrighted material by default. In March 2026 the government finally published its statutory report – and declined to change anything at all. [6]
That sounds like a fudge, and in fairness it largely is. But the practical effect is significant: copying protected works for commercial AI training in the UK still requires a licence. [7] No new exception, no opt-out mechanism, no free pass. Developers who’ve already hoovered up protected works without permission are sitting on unresolved exposure, and the government has promised a voluntary licensing code and transparency obligations without publishing a draft or a timetable for either.
Translation: the legal position hasn’t improved for AI companies. It’s just stopped getting worse for the rest of us. I’ll take it.
Here’s where it gets particularly sneaky. Suno offers “ownership” of generated tracks to paid subscribers. Sounds great, doesn’t it? Except their own help documentation explicitly admits this may not include copyright protection. [8]
| What Suno Claims to Offer | What You Actually Get | What You Need |
|---|---|---|
| “Ownership” of tracks | Possibly ownership of the file | Copyright protection |
| Commercial use rights | Permission to use commercially | Legal defensibility if challenged |
| “Keep your copyrights” (Pro/Premier) | Unknown – company admits uncertainty | Clear, established copyright |
| AI-generated music | Public domain output (per US law) | Protected intellectual property |
Suno’s own documentation states: “In the US, copyright laws protect material created by a human. Music made 100% with AI would not qualify for copyright protection because a human did not write the lyrics or the music.” [8]
It’s like being given keys to a car whilst being told you might not actually be allowed to drive it. The keys are lovely, very shiny, but rather miss the point.
As you don’t own the copyright, there’s absolutely no way to stop someone from claiming it belongs to them, uploading, and distributing the track, and registering it with fingerprinting services such as ContentID. Next thing you know, every single video gets a copyright copyright claim, and all revenue is suddenly, and automatically directed to them while you fight the bots and get nowhere. Or worse these rogue parties register a full strike with you and your channel gets suspended and possibly deleted.
Don’t ask me how I know…
Now, you might be thinking this is all theoretical hand-wringing. I’m afraid not. The lawsuits are already here, they’re massive, and they’re going to reshape the entire landscape.
In June 2024, all three major music labels – Universal, Sony, and Warner – launched coordinated lawsuits against both Suno and Udio through the Recording Industry Association of America (RIAA). [9][10] This isn’t a small dispute over a few tracks. The RIAA accused these companies of “mass infringement of copyrighted sound recordings on an almost unimaginable scale.” [9]
Here’s the genuinely alarming bit: Suno admitted to using copyrighted music for training their AI models. [11][12] They’re arguing it’s “fair use,” but that defence is far from established in this context. The potential damages? Up to $150,000 per infringed track. [13]
These cases are ongoing in US District Courts as I write this. The outcomes will likely determine the future of AI music generation.
By November 2025, we’re seeing a pattern emerge. Udio settled with both Universal Music Group and Warner Music, though the terms remain confidential. [14] The labels are now pivoting to a “walled garden” model – licensing their catalogues to AI companies under strict conditions.
And just to put the scale of these legal battles in perspective: in September 2025, a group of authors settled a copyright infringement case against AI company Anthropic for $1.5 billion over the use of pirated books to train AI models. [15] That’s billion with a B. The largest AI copyright settlement ever recorded.
Legal experts predict more private settlements with AI companies offering equity stakes to rights holders rather than admission of wrongdoing. [15] Translation: the big players are paying enormous sums to make these problems go away quietly.
If the platforms creating AI music are admitting to training on unlicensed music, and major labels are suing for infringement with potential billion-dollar damages, where exactly does that leave content creators who innocently used that AI music?
Spoiler: potentially holding a very large, very expensive bag.
UPDATE (January 29, 2026): News has broken that illustrates just how serious the music industry is about AI copyright. Universal Music Group, Concord, and ABKCO have filed what could be “the single largest non-class action copyright case in US history” – suing an AI company for over $3 billion over alleged infringement of more than 20,000 songs.[28]
This isn’t about AI music generators like Suno or Udio. This lawsuit targets an AI text company whose models allegedly ingested copyrighted music lyrics without permission. The complaint claims the company’s founders personally torrented 5 million pirated books (including songbooks) from notorious pirate site LibGen, despite internally describing the source as “sketchy” and a “blatant violation of copyright.”[28]
The message from the music industry couldn’t be clearer: they’re going after AI companies with unprecedented legal force, regardless of whether those companies generate music or just use music in their training data. If you’re a content creator relying on AI-generated anything that might have touched copyrighted music, you’re standing uncomfortably close to a legal battlefield measured in billions.
Since first publishing this update: the case has escalated rather than settled. The publishers filed for partial summary judgment in their original 2023 case in March 2026, arguing the evidence is no longer meaningfully contested, and amended that complaint again in July 2026 using material obtained in discovery. The separate $3bn suit heads into motion-to-dismiss briefing in August 2026. Meanwhile the $1.5 billion settlement with authors – the one I mentioned above – has now received court approval. [40] These things do not appear to be going away.
UPDATE 2 (March 19, 2026): And the tide keeps turning. Yesterday, the UK government announced it’s scrapping plans that would have allowed AI companies to train their models on copyrighted music without explicit permission.[29] The original proposal – which would have made unauthorised use of copyrighted material the default, with creators having to actively “opt-out” – sparked massive backlash from the music industry.
Of 11,520 consultation submissions, 81% named mandatory licensing as their preferred approach. Just 3% supported the government’s own “preferred” option.[30] Major artists including Elton John, Dua Lipa, Björn Ulvaeus (ABBA), Max Richter, and Thom Yorke (Radiohead) all campaigned against what Sir Elton described as “committing theft, thievery on a high scale.”[29]
UK Culture Secretary Liz Kendall confirmed the U-turn, stating that “copyright material cannot be used for AI development and training without permission.”[29] UK Music called it “a major victory for campaigners” and emphasised that the hundreds of thousands working in the industry should be able to work “without the constant fear that the fruits of their labour could effectively be taken by AI firms without payment or permission.”[29]
This marks a significant policy shift: governments are starting to side with creators over AI companies. The legal and regulatory landscape isn’t just uncertain – it’s actively moving against unlicensed AI training. Worth considering before you commit your content strategy to AI-generated music, isn’t it?
UPDATE 3 (31st July 2026): Right, this is the big one. A Munich court has just ruled that Suno infringed copyright.
Germany’s collecting society GEMA sued Suno in January 2025, arguing that the tool’s outputs were reproducing works it represents. Today Judge Elke Schwager agreed. Suno wasn’t entitled to use music from GEMA’s songwriters and composers, must disclose the revenue it earned doing so, and faces damages yet to be quantified. [31] GEMA’s chief executive called it a verdict of global significance, which is the sort of thing you’d expect a winning CEO to say, except in this instance he’s probably right. Suno disagrees with the ruling and is weighing an appeal. [32]
Two things make this bigger than a single court case.
First, it’s the first major European ruling to tell AI companies that operating without an explicit, opt-in publishing licence carries real financial consequences. [33] Not “might carry”. Does.
Second, it’s the same judge who ruled against OpenAI in November 2025 for reproducing German song lyrics. [34] Munich is quietly becoming the venue where “the model just learned patterns, it didn’t copy anything” stops working as an argument.
What it doesn’t do is settle anything in the US. A German ruling sets no precedent in Massachusetts, where Sony and Universal are still litigating against Suno. Warner settled in November 2025 and took a licensing deal instead. Sony and Universal didn’t, and in May 2026 they moved to expand the case from 560 recordings to over 61,000 – a move that would push potential statutory damages past $9 billion. [35]
So where does that leave you, sitting there with a back catalogue of AI-generated background music? Roughly where you were yesterday, but with rather less room to argue that this is all speculative hand-wringing from a bitter composer. A court has now looked at it and said: no, this was infringement.
One final twist, and it’s a strange one. In July 2026 the American Federation of Musicians sued Universal and Warner – not Suno, not Udio, but the labels themselves. [41] The union alleges that session recordings made by its members were licensed to the AI companies for training without the consent or compensation their collective bargaining agreement requires.
Sit with that for a moment. The “clean” licensed AI models – the ones built on properly negotiated deals, the ones we’re all being told represent the responsible path forward – may themselves be sitting on rights that were never actually cleared with the humans who played the notes.
Which rather illustrates the whole problem, doesn’t it? At every layer of this, somebody’s assuming somebody else did the paperwork. When you use my music, there is precisely one person in that chain, and he’s writing this sentence.
Right, let’s talk about what happens when – not if, when – the legal dust settles. There are several possible outcomes, and none of them are particularly reassuring for anyone who’s built their content strategy around AI music.
If courts definitively rule that training AI on copyrighted music without licensing is infringement (which seems increasingly likely given the admission and settlement patterns), AI music companies face either massive damages or complete shutdown.
What happens to all the music generated on those platforms? The legal status becomes even murkier. Your back catalogue of AI music? Suddenly radioactive. No platform wants to host content with unclear legal provenance, and you’re left explaining to YouTube why that copyright claim just arrived.
If courts affirm the Copyright Office’s position that AI-generated music can’t be copyrighted, that creates its own problems. Your AI music can’t be protected – anyone can copy it, reuse it, or claim it as their own.
Found the perfect AI-generated track for your brand? Congratulations, so can your competitor. Someone steals your AI track? No legal recourse whatsoever.
Legal experts at Clifford Chance noted the paradox: “If AI-generated music does not attract copyright protection, it is in a better position to undercut human-made copyright music.” [16] Your unprotected AI music competing with other people’s unprotected AI music – it’s a race to the bottom.
This one isn’t hypothetical – it’s already underway, and it’s moved fast since I first wrote this.
YouTube updated its policies in July 2025 to address AI-generated content. Music without “clear human input” may face limited reach, blocked monetisation, or takedowns. [17]
The streaming platforms have gone considerably further. Spotify adopted the DDEX standard for AI disclosure in track credits in September 2025, and began testing listener-visible AI Credits in April 2026. [36] Apple Music made Transparency Tags a delivery requirement from March 2026, covering artwork, audio, composition and music video, with listener-facing labels rolling out from May. [37] Deezer has been the most aggressive: it built its own detection system, has tagged more than 13 million AI tracks, and keeps flagged material out of editorial playlists and algorithmic recommendations entirely. [38]
The numbers behind that crackdown are genuinely startling. Deezer reported that by late 2025 roughly half of all new uploads contained AI-generated audio, and that 85% of streams on AI-generated music were detected as fraudulent. [38] Spotify removed around 75 million spam tracks in twelve months. [19] A Deezer-Ipsos survey found that 80% of listeners think AI music should be labelled – and that 97% couldn’t tell the difference without being told. [38]
And then there’s the bit with an actual deadline attached. Article 50 of the EU AI Act becomes enforceable on 2nd August 2026, requiring machine-readable marking of AI-generated audio. [39] Not a platform policy you can shop around. A law.
Here’s what that means in practice. The disclosure metadata your distributor sends downstream is about to become the thing that determines how your music is treated – whether it gets recommended, whether it gets monetised, whether it stays up. Get it wrong in either direction and you’ve got a problem: undisclosed AI risks removal, and human-made tracks with sloppy metadata are increasingly getting caught in filters built for AI.
The era of quietly uploading AI music and hoping nobody notices is closing. Rather sharply.
Here’s what keeps me up at night on your behalf. When something goes wrong with AI music, who takes responsibility?
The AI company’s terms and conditions (you did read those, didn’t you?) shift liability to users. If a platform removes your content, if you face a copyright claim, if legal questions arise – who do you call?
With AI music: no accountability, no support, no human to help. Your channel hangs in the balance whilst lawyers argue about whether your background music even exists in a legal sense.
Rather less appealing now, isn’t it?
If you’re thinking this might blow over, let me disabuse you of that notion. The music industry has seen what happened when it initially ignored digital disruption, and it’s not making that mistake twice.
RIAA Chairman Mitch Glazier didn’t mince words: “Unlicensed services like Suno and Udio that claim it’s ‘fair’ to copy an artist’s life’s work and exploit it for their own profit without consent or pay set back the promise of genuinely innovative AI for us all.” [9]
All three major labels have filed lawsuits and simultaneously pivoted to licensing deals with AI companies willing to play by the rules. The message is clear: work with us or face legal oblivion.
In April 2024, over 200 prominent artists signed an open letter specifically targeting AI music generators. [20][21] The signatories include Billie Eilish, Nicki Minaj, Stevie Wonder, Katy Perry, and the estates of Bob Marley and Frank Sinatra.
When Billie Eilish and Frank Sinatra’s estate agree on something, it’s probably worth paying attention. (Not a sentence I ever expected to write, but here we are.)
The letter warned against what they called “this assault on human creativity” and the incoming “deluge of AI-generated noise.” [20] These aren’t fringe voices – these are the artists who shape popular culture and whose work you’ve likely enjoyed for years.
Remember the AI-generated Drake and Weeknd collaboration that went viral in 2023? It racked up over 9 million views before Universal Music Group filed takedowns across every platform. [22] The song was also ruled ineligible for Grammy consideration. [23]
The lesson: major labels have the resources, the legal teams, and the determination to pursue AI content that infringes on their artists’ rights. They will win, and you don’t want to be caught in the crossfire.
Let’s make this personal. You’re not a legal scholar or a music industry executive. You’re a content creator trying to make engaging videos without spending your entire budget on music licensing. I get it – that’s precisely why I offer free music with Creative Commons licences.
But here’s what’s actually happening to creators using AI music.
Creators are already receiving copyright claims on videos using AI-generated music. One YouTuber described the process: “I submitted a dispute but the company has rejected my dispute. My next option would be to appeal the decision but if my appeal is also rejected, I will receive a copyright strike.” [24]
Think about that for a moment. You used “free” AI music, and now you’re facing a potential copyright strike that could permanently damage your channel. Three strikes, and your channel gets terminated. Permanently.
When something goes wrong with your music, who helps you?
The difference is stark. AI music is trained on copyrighted songs. If the output resembles anything in that training data – and given how training works, it almost certainly does – you might face infringement claims. And you’ll be facing them alone.
Legal expert Victoria Hasselman warned: “Artists will simply lie about the role of the AI when they apply to the copyright office. Before you know it, we’ll have thousands of registrations and no visible way of checking.” [25] Translation: the copyright database itself may become unreliable, making your position even more precarious.
You’ve spent hours crafting the perfect video. The editing is flawless, the pacing is spot-on, you’ve nailed the storytelling. Then comes the copyright claim notification at 2 AM.
Is saving 10-30 minutes searching for a perfectly fitting track, or £20 on music licensing worth risking all that work? Worth potentially losing your channel? Worth the stress of fighting claims you don’t understand?
I’d argue not, but then again, I’m a dinosaur human composer with a vested interest in your answer.
Right, enough doom and gloom. Let me tell you why human-created music is still the sensible choice. (Turns out there are some advantages to being a dinosaur.)
Human-created music benefits from centuries of legal precedent. Copyright ownership is unambiguous. Your rights are clearly defined and defensible in court should it ever come to that.
When I create a track, I own it. Full stop. No asterisks, no “maybe,” no legal grey areas. When you use my music under Creative Commons, those rights are internationally recognised and tested through decades of use.
Unlike an AI, I can explain exactly how each track was made. I can personally vouch for its originality. If you run into any problems – copyright claims, licensing questions, platform issues – there’s a real human (me) who answers emails and actually wants to help.
Every track on this site was created in my studio, by me, with my own hands. There’s a clear chain of custody from creation to your download.
Creative Commons licences, particularly CC BY 4.0 (which I use), are internationally recognised legal frameworks. [26] Millions of creators have used CC-licensed content for years without issue. The legal ground is solid, tested, and reliable.
Compare that to AI music’s “maybe you’re protected, possibly you’re not, ask your lawyer” approach.
Here’s what you get with human-composed royalty-free music: you download the track, you credit the creator (if required by the licence), and you get on with making your content. Done.
No worrying about whether your music will legally exist next year. No wondering if copyright law will shift underneath you. No concerns about platform policy changes leaving you stranded.
Legal expert analysis confirms: “An artist who uses AI to assist and enhance their own creativity is unlikely to limit copyright protection for their music as long as they avoid allowing the AI to make expressive choices for them.” [27] Translation: human involvement equals legal safety.
It’s almost as if centuries of copyright law were designed around, you know, humans creating things.
Look, I’m not here to judge anyone who’s already used AI music. Those tools are genuinely appealing, and the marketing makes them sound perfectly safe. But now that you know what you’re dealing with, here’s some practical advice.
(AGAIN Remember: I’m not a lawyer, and this isn’t legal advice. I’m a composer who wants content creators to make informed choices.)
Document everything. Save your prompts, save your creative process, keep records of any modifications you made. If you end up in a legal dispute, this documentation might be your only defence.
Add significant human modifications. The more you genuinely transform the AI output, the stronger your legal position. This isn’t just tweaking the EQ – this means substantial creative changes.
Write your own lyrics if the track includes them (no, not AI generated lyrics…) Lyrics CAN be copyrighted, even if the underlying music cannot. [8] At least you’ll have some protectable elements.
Avoid prompts referencing specific artists or songs. “Make something like Drake” or “in the style of Taylor Swift” is a legal minefield waiting to explode.
The US Copyright Office recommends fully disclosing any AI-generated portions when registering work. [2] Legal advisers suggest that human “selection, arrangement, and creative modifications may qualify” for copyright protection. [27]
In other words: use AI as a tool to assist you, not as a replacement for human creativity. The more human involvement, the better your legal position.
What happens if the AI platform you’re using shuts down or changes its terms? Can you prove the provenance of your music if challenged in three years? Is the short-term convenience worth the long-term uncertainty?
These are questions worth asking before you commit your entire back catalogue to legally dubious AI music.
Right, I’ve thrown a lot of information at you. Let’s bring this back to earth.
AI music generators aren’t inherently evil – they’re genuinely impressive technology created by clever people. The problem isn’t the technology itself; it’s the legal foundation (or lack thereof) underneath it.
The legal status of AI-generated music ranges from uncertain to nonexistent. Major litigation is ongoing. The landscape could shift dramatically at any moment. When that happens – and it will happen – content creators using AI music may find themselves legally exposed with no support network.
Here’s what I believe: free music shouldn’t mean you’re on your own when legal issues arise. It shouldn’t mean gambling with your channel’s future. It shouldn’t mean lying awake wondering if your background music might cause your content to be taken down.
Clear copyright, accountability, human support, and legal certainty – these aren’t luxuries. They’re necessities for building a sustainable content creation career.
If you’re looking for music that won’t land you in legal hot water, you’re in the right place. Every track in my library was composed by an actual human (me), comes with clear Creative Commons licensing, and includes personal support if you ever need it.
Browse the collection, find something that fits your project, and create your content with confidence. And if you ever have questions about using my music – licensing, copyright, technical issues, whether my cat deserves a credit – just drop me a line. I’m a real human, and I actually respond.
After all, that’s one thing AI definitely can’t replicate.
Shane Ivers is a composer and founder of Silverman Sound Studios, providing free royalty-free music to content creators worldwide. With a Master’s degree in Music specialising in Electronic Composition and over 20 years of professional experience, Shane creates free-range organic music across diverse genres under the Creative Commons CCBY 4.0 licence. Unlike AI music generators, Shane personally stands behind every track and offers direct support to creators using his music.
[1] Suno Terms of Service, “Copyright and Ownership” section, Suno AI. https://suno.com/terms
[2] U.S. Copyright Office, “Copyright Office Releases Part 2 of Artificial Intelligence Report” (January 29, 2025). https://www.copyright.gov/newsnet/2025/1060.html
[3] Music 3.0, “5 Things The US Copyright Office Decided About AI-Generated Content” (February 2025). https://music3point0.com/2025/02/05/5-things-the-us-copyright-office-decided-about-ai-generated-content/
[4] Thaler v. Perlmutter, affirmed 2025, establishing that copyright protection is reserved for works of human creation.
[5] UK Copyright, Designs and Patents Act 1988, Section 9(3), regarding computer-generated works. https://www.aoshearman.com/en/insights/ownership-of-ai-generated-content-in-the-uk
[6] UK Intellectual Property Office consultation on AI and copyright. https://www.lexology.com/library/detail.aspx?g=616b1fdd-bc88-444d-823e-388ba77e8339
[7] UK Government AI Economic Impact Report, expected 2026. https://www.finnegan.com/en/insights/articles/mind-the-copyright-the-uks-ai-and-copyright-conundrum.html
[8] Suno Help Center, “Do I have the copyrights to songs I made?” https://help.suno.com/en/articles/2746945
[9] Recording Industry Association of America, “Record Companies Bring Landmark Cases for Responsible AI Against Suno and Udio” (June 2024). https://www.riaa.com/record-companies-bring-landmark-cases-for-responsible-ai-againstsuno-and-udio-in-boston-and-new-york-federal-courts-respectively/
[10] CNBC, “Music labels sue AI companies Suno, Udio for U.S. copyright infringement” (June 24, 2024). https://www.cnbc.com/2024/06/24/music-labels-sue-ai-companies-suno-udio-for-us-copyright-infringement.html
[11] TechCrunch, “AI music startup Suno claims training model on copyrighted music is ‘fair use'” (August 1, 2024). https://techcrunch.com/2024/08/01/ai-music-startup-suno-response-riaa-lawsuit/
[12] Deadline, “AI Startup Suno Claims ‘Fair Use’ Copyright Doctrine Allows Training On Major Recordings” (August 2024). https://deadline.com/2024/08/ai-startup-suno-claims-copyright-doctrine-allows-ai-training-1236029563/
[13] Crowell & Moring, “Major American Music Labels Sue Generative AI Music Platforms in First Case of Its Kind Over AI Audio.” https://www.crowell.com/en/insights/client-alerts/major-american-music-labels-sue-generative-ai-music-platforms-in-first-case-of-its-kind-over-ai-audio
[14] TechCrunch, “Warner Music settles copyright lawsuit with Udio, signs deal for AI music platform” (November 19, 2025). https://techcrunch.com/2025/11/19/warner-music-settles-copyright-lawsuit-with-udio-signs-deal-for-ai-music-platform/
[15] Medium, “Using Suno.AI Legally: A Guide to Copyright and AI-Generated Music in 2025” (discussion of major settlements including Bartz v. Anthropic). https://medium.com/@wackyworld_jenshenneberg/using-suno-ai-legally-a-guide-to-copyright-and-ai-generated-music-in-2025-b4ddd77b4bce
[16] Clifford Chance legal analysis on AI-generated music copyright paradox, cited in multiple legal reviews.
[17] YouTube AI content policy updates, July 2025, regarding music without clear human input.
[18] Deezer reports on AI-generated music submissions, 2024-2025.
[19] Spotify removal of AI-generated spam tracks, reported across multiple industry sources 2024-2025.
[20] TechCrunch, “Nicki Minaj, Billie Eilish, Katy Perry and other musicians sign letter against irresponsible AI” (April 2, 2024). https://techcrunch.com/2024/04/02/nicki-minaj-billie-eilish-katy-perry-and-other-musicians-sign-letter-against-irresponsible-ai/
[21] Billboard, “Open Letter on AI Music Signed By Billie Eilish, Pearl Jam, Nicki Minaj” (April 2024). https://www.billboard.com/business/tech/open-letter-ai-music-signed-billie-eilish-pearl-jam-nicki-minaj-1235647311/
[22] Variety and multiple sources on the “Heart on My Sleeve” AI Drake/Weeknd incident and subsequent takedowns (2023).
[23] Grammy eligibility ruling on AI-generated “Heart on My Sleeve,” 2023.
[24] Creator testimony from various YouTube and creator forums regarding AI music copyright claims, 2024-2025.
[25] Victoria Hasselman, legal expert, quoted in Jenner & Block analysis: “AI Music Raises Fresh Copyright Issues for Lawyers.” https://www.jenner.com/en/news-insights/publications/ai-music-raises-fresh-copyright-issues-for-lawyers-daily-journal
[26] Creative Commons licensing framework documentation, Creative Commons International.
[27] Rimon Law, “How Copyright Office Guidance Applies to Music That Includes AI-generated Material.” https://www.rimonlaw.com/how-copyright-office-guidance-applies-to-music-that-includes-ai-generated-material/
[28] Music Business Worldwide, “UMG sues Anthropic for $3bn over ‘brazen’ copyright infringement of 20,000+ songs” (January 28, 2026). https://www.musicbusinessworldwide.com/umg-concord-and-abkco-sue-anthropic-for-3bn-in-what-could-be-single-largest-non-class-action-copyright-case-in-us-history/
[29] BBC News / Multiple sources, “UK government abandons plans to give AI models access to copyrighted music” (March 18, 2026). Government confirms copyright material cannot be used for AI training without permission following widespread industry backlash. https://www.bbc.co.uk/news/articles/cvg1gr5v333o
[30] Complete Music Update, “Music and AI: 2025’s developments that will shape 2026’s disputes” (January 2026). 11,520 consultation submissions; 81% named mandatory licensing as their preferred approach, 3% backed the government’s opt-out proposal. https://completemusicupdate.com/music-and-ai-2025s-developments-that-will-shape-2026s-disputes/
[31] Reuters, “German court rules AI music firm Suno broke copyright rules” (31st July 2026).
[32] Music Business Worldwide, “Suno loses copyright infringement lawsuit brought by GEMA in Germany” (31st July 2026). https://www.musicbusinessworldwide.com/suno-infringed-copyright-in-gema-case-german-court-rules/
[33] Music Ally, “German collecting society GEMA wins its copyright-infringement lawsuit against Suno” (31st July 2026). https://musically.com/2026/07/31/german-collecting-society-gema-wins-its-copyright-infringement-lawsuit-against-suno/
[34] Reuters, Munich Regional Court ruling against OpenAI over reproduction of German song lyrics (November 2025).
[35] Sony Music and Universal Music Group motion to expand the Suno case to 61,026 recordings, D. Mass. (May 2026).
[36] Spotify AI disclosure policy (September 2025) and AI Credits beta (April 2026), reported across industry sources.
[37] Apple Music Transparency Tags, announced March 2026 as a delivery requirement, listener-facing rollout from May 2026.
[38] Deezer AI detection and tagging data, 2025–2026, including Deezer-Ipsos listener survey.
[39] EU Artificial Intelligence Act, Article 50, transparency obligations for AI-generated content, enforceable from 2nd August 2026.
[40] Music Business Worldwide, “Music publishers file amended lyrics lawsuit against Anthropic – just as AI firm’s separate $1.5B piracy settlement with authors wins court approval” (July 2026).
[41] American Federation of Musicians amended complaint against Universal Music Group and Warner Music Group (July 2026).
Additional Resources:
Did you find this article helpful? Share it with fellow content creators who might be considering AI music for their projects.
Have questions about copyright, licensing, or using royalty-free music? Contact Shane directly – unlike AI, he actually responds.
You’ve spent hours crafting the perfect holiday video. The editing’s spot-on, the content’s brilliant, and you’ve queued up “Silent Night” as background music. Surely that’s public domain, right? It’s from the 1800s, after all.
I think we can all agree that music enhances a game or film. Music gives meaning, mood, continuity, memorability, and absorption.
What if I told you that you can get professional quality music, that is created from scratch, is formulated to fit your project perfectly, and can even feature a real full orchestral score, for less than you think?
Don’t let anything stop you from writing that song! I’ve gathered my best and most practical processes to break down the walls of songwriter’s block, and get making sweet music once more!
Wow, you made it here? It seems that you like what you hear! Fancy receiving some musical goodness directly into your mailbox?
Feel amazing in 60 seconds time you generous, wonderful person…
Become a patron of the musical arts – feel cool, and get rewarded…
Become a musical trendsetter, and never miss a track…
14 responses
A VERY interesting read. I (we, band members) recently used AI to produce a new version of an song we wrote and recorded back in the 80s and were blown away by how it handled it.
It was specifically instructed to adhere to the melody and lyrics and chord progressions supplied in the original upload etc, plus, our original version from the 80s was previously published on YouTube, uploaded about 15 years ago, which it can be directly compared to.
Copyright *should* not be an issue this instance.
But as regards using AI to create a song, a much different kettle of fish, as you’ve outlined.
That’s a genuinely interesting use case — and you’re right that the copyright situation there is fundamentally different. If you’re the original author, providing your own melody, lyrics, and chord progressions, and using AI essentially as a production or arrangement tool rather than a creative one, you’ve retained the human authorship that copyright law actually cares about. The fact that you have an original 1980s YouTube upload as a verifiable reference doesn’t hurt either. It’s the “type a prompt and see what pops out” approach where things get legally murky — what you’ve described is much closer to using a skilled arranger than to generating something from thin air. Sounds like it came out well, too! Shane
I write lyrics but I don’t play in a band. The most of a musical instrument I have held was a triangle at school. I have tried sending my lyrics to band/ artist. But they don’t get back to you. So when I tried an AI to do it I was impressed. So I have wrote the lyrics but the AI has done the music. I carnt copyright the song. Then the music company’s should make it easier for people to get there lyrics heard.
That’s a genuinely frustrating situation, and I think you’ve put your finger on a real problem — the traditional routes for getting lyrics in front of artists are almost entirely closed to people without connections. It’s not a reflection of the quality of the writing; it’s just how the industry is structured.
The copyright position for lyrics is actually more favourable than it is for the AI-generated music underneath them. Under current US Copyright Office guidance, lyrics written by a human can be copyrighted even if the accompanying music can’t — so your words do have protection, even if the track as a whole doesn’t. That’s worth knowing, and worth documenting properly.
On finding musicians to collaborate with, platforms like Kompoz or SoundBetter are designed specifically for exactly this kind of remote collaboration, if you haven’t tried those already.
Shane
Hi Gaz, I understand the problem . I write songs eg i have put music to help a friend who writes a bit of poetry type lyrics who didn’t mind me changing or adding on some different words to sing easier.
This may not suit you if you need to not change any words but if you don’t mind some word adjustments you are welcome to send me to view
Kind regards
Nic Rado
I’ve gone to some lengths to decipher the music AI ‘mentality’. As the creator of this blog mentioned, it’s how clever you’re capable of being. The short answer? Yes, you can ‘get away with it’. The long and safest answer is, upload your original bare-bones full composition and lyrics. The likes of Suno will polish it (make it more attractive) and may add some extra instrumental/vocals fills here and there. Use the preferred version Suno created and record the whole thing with real humans/computer/paid VST libraries, including lead singer/BV’s (the latter if needed). But before you do record, pay attention to the ‘extra fills’ that Suno included and make tweaks so it does not end up being exactly the same. It is of course safer when you are a paid subscriber. No one, not even Suno, will have a leg to stand on.
Option 2 is you use a song entirely created by Suno, or similar (but still with your own lyrics) and make such heavy changes that no one that listens to both versions will say they are exactly the same or not different enough. Many modern ‘pop’ songs have similar chord structures and there is no way any artist will sue the other for having similar chord structures.
Remember that before AI many artists copied some stuff from others. It’s the nature of things.
The workflow you’ve outlined in option one is essentially what the Copyright Office guidance points toward as the legally defensible approach — using AI as a tool to assist human creativity rather than replacing it. If you’re supplying the core composition and only using Suno to embellish or polish it, there’s a reasonable argument that the human authorship is still intact, particularly if you then re-record the whole thing with real instruments and make deliberate changes to anything AI-added.
Where I’d still urge caution is documentation. If you ever needed to defend your authorship — whether against a copyright claim or to register the work — being able to show the original composition, the Suno session, and the subsequent human modifications would be essential. “I made enough changes” is a much harder argument to win without a paper trail.
Option two is where I’d pump the brakes a bit, personally. “Heavy enough modifications that listeners can’t tell” is a subjective standard that tends to look a lot shakier when a lawyer’s involved. But you’re right that chord progressions alone aren’t protectable — it’s more about melodic and rhythmic similarity, and even that gets genuinely complicated.
Shane
Je m’amuse à faire ça aussi, ou d’utiliser le mastering et le mixage en n’oubliant jamais de décocher la case “autorise le remixage de ma musique”, (j’ai déjà retrouvé un de mes morceaux reutilisé avec un texte à la con et un arrangement très commun sur Suno, qui l’a quand même fait sauter après mon signalement) l’IA doit être utilisée comme un séquenceur, ou une station RaW complète, mais pas pour de la composition. Je crois que qu’elles commencent à évoluer par là. .. Heureusement. Sam Chökk (bandcamp, Spotify, youtube,…)
Good to hear the takedown worked — that’s unfortunately exactly the kind of thing I was alluding to in the article with the “don’t ask me how I know” line. Finding your own work turned into something with bargain-bin lyrics slapped over it is a particular kind of grim.
The sequencer analogy is spot on, and I think it’s the right frame for where this should land. Nobody argued that DAWs or sample libraries were going to kill composition — they just became part of the toolkit. The difference is those tools don’t make creative decisions for you; they execute them. That’s the line AI keeps trying to blur, and you’re right that there are signs of movement toward a more tool-like model, even if it’s being dragged there kicking and screaming by litigation rather than by any great philosophical awakening on the part of the companies involved.
Shane
If AI music cannot be copyrigthed in the US, why is YouTube supporting AI music against the real musician of whom the AI was trained with and then claims the real musician to be illegally copying the music and taking all the money, even from older songs?
YouTube makes no proof whatsoever and gives a shit about what’s real and what’s not.
Just claim “it is stolen from me” first and YouTube will work for you. Bunch of As….es
YouTube’s ContentID system was already open to massive abuse, and that’s before AI music arrived on the scene. Now the system is being presented with all these new AI tracks, and it can’t distinguish between human and AI. Unless they totally overhaul the system I don’t know how it’s going to get anything but worse. Great point!
Shane
This is a massive wake-up call for creators who are leaning too heavily on AI for their core assets. While it’s fun to experiment with things like Arthur Morgan AI speech for quick social clips, the lack of copyright protection for AI music is a genuine legal minefield. I’ll definitely be sticking to human composers for anything commercial after reading this.
I have 300+ songs I wrote and provided a prompt. Never before were they produced. I ran them through SUNO. Where do I stand with copyright, use, promote?
Sorry for the slow reply, Paul — and it’s a good question, because the answer genuinely depends on something your comment doesn’t quite specify. Usual caveat: I’m a composer, not a lawyer, and none of this is legal advice — with a catalogue that size it’s genuinely worth an hour of a proper IP solicitor’s time.
If you wrote lyrics and then prompted Suno to build music around them, you own the lyrics and only the lyrics. Human-written words are copyrightable; the generated audio underneath them, under current US Copyright Office guidance, is not. If you wrote actual songs — melody, chords, structure — and used Suno to produce and arrange them, you’re in a much stronger position, because the composition is yours and Suno was doing a producer’s job rather than a writer’s. That distinction is the whole ballgame, and it’s worth being honest with yourself about which one you did.
Either way, three practical things.
Document what you’ve got, now. Original lyric sheets, dates, any demos or voice memos, the prompts themselves. If you ever need to prove authorship — to register the work, to fight a claim, to license it — “I definitely wrote this” is a much weaker argument than a timestamped paper trail across 300 songs.
Be careful about promoting and distributing. Every major platform now requires AI disclosure through your distributor, and the EU’s rules on marking AI-generated audio become enforceable in August. Declare it accurately. Undisclosed AI is now the thing that gets releases pulled and accounts reviewed, and with a catalogue that size, one bad flag can jeopardise the lot. Also worth knowing: you almost certainly can’t register these with Content ID, and you’d be unwise to try.
Understand what you can’t stop. Where the music isn’t protected, you have no legal route to prevent someone else uploading it, distributing it, or claiming it. That’s not a hypothetical — it’s the single most common way people get burned here.
If these songs matter to you, the strongest move by some distance is to re-record the ones you care about most with real musicians, using the Suno versions as demos. Expensive and slow, yes. But it converts an unprotectable file into a work you actually own. Three hundred is a lot to be holding on uncertain ground.
Shane