In 2023, Ed Sheeran stood in a New York courtroom, guitar in hand, playing "Thinking Out Loud" to prove he hadn't stolen from Marvin Gaye's "Let's Get It On." This was a perfect snapshot of our current musical situation, where copyright disputes are surging and artists repeatedly face that annoying question: who owns a melody?
The music world has gotten quite chaotic when it comes to who stole what from whom and the outcomes of these legal battles can be expensive. When Robin Thicke and Pharrell Williams lost their "Blurred Lines" case, they were hit with a $7.3 million jury award (later reduced to $5.3 million) for allegedly copying Marvin Gaye's "Got To Give It Up." That verdict definitely made musicians and their lawyers nervous about what exactly constitutes theft.
And with all AI tools that can generate a "new" Taylor Swift song in seconds, popping out every day, the lines between inspiration, imitation, and infringement have become so blurry that even the courts are struggling to keep up, creating an entirely new category of potential legal nightmares.
Read also: Copyright Challenges in AI-Generated Music: Who Really Owns the Melody?
With all that, what actually counts as music plagiarism? How do you prove someone stole your song? And what happens when the "thief" isn't even human? In this article, we'll try to get into the messy world of musical plagiarism and copyright infringement. Because in a world where anyone (or anything) can create a song that sounds just like yours, knowing how to protect your work is getting more important even as you read this article.
The Anatomy of Musical Theft: What Actually Constitutes Plagiarism?
When musicians get accused of stealing someone else's song, two terms get thrown around quite a bit — plagiarism and copyright infringement. And while people use them like they mean the same thing, they don't.
Plagiarism in music means taking someone's work and pretending it's yours, which many consider unethical but isn't actually against the law. Copyright infringement, however, happens when someone uses protected parts of a song without permission. So you can plagiarize without breaking copyright law, and you can infringe copyright without meaning to plagiarize.
"Plagiarism is not a legal term. When plagiarism occurs, the legal question is whether it was a copyright infringement," explains Matthew Asbell, a partner at Lippes Mathias' intellectual property team. "The courts and lawyers have a somewhat convoluted way of evaluating when there is infringement. Copyright infringement occurs when the music is sufficiently expressive (i.e. creative) and fixed (i.e., recorded or notated) such that it is protected by copyright, when the defendant had access to the work, when the defendant copied, distributed, modified, or performed music that is 'substantially similar' to the protected music, and when the defendant's action was not excused as a 'fair use'."
This "substantial similarity" test is quite central to how courts decide these cases. When a songwriter sues, they need to prove two things: the alleged thief had access to the original song, and the new song is substantially similar to protected elements of the original.
And while that sounds straightforward enough, it gets complicated. Courts have to filter out basic musical elements that nobody can own.
Mark S. Lee, Partner at Rimon Law who advises high-level celebrities and studios, puts it this way: "Copyright infringement (correct legal term for 'plagiarism') takes place when someone copies what is called 'protectable expression,' as opposed to unoriginal ideas, without the permission of the copyright owner. Copying original melodies, lyrics, etc. can be infringing; copying a chord progression usually would not be, unless the precise way a progression was expressed in an underlying song was also copied."
The Ed Sheeran case shows how this works in practice. The jury sided with Sheeran, agreeing that similarities between the songs were just basic musical building blocks that nobody can claim ownership over.
Asbell elaborates on why Sheeran won: "This is why Ed Sheeran's 'Thinking Out Loud' was not deemed to be an infringement of Marvin Gaye's 'Let's Get It On'. Even though there are significant similarities, the elements that were allegedly copied were considered fundamental musical building blocks that are not subject to copyright protection. You can think of how the letters that make up this response are not capable of being owned and controlled exclusively because people need to be able to use them to form other sentences, but the words in order that form the sentences and paragraphs are protectable, and thus could be asserted successfully if they were copied without authorization."
But some cases are much more clear-cut. "One of my favorites, perhaps because of my partiality to Brazilian music, was a suit filed by Jorge Ben Jor and Gilberto Gil against Rod Stewart for copying the very distinctive melody of 'Taj Mahal' in his song 'Do You Think I'm Sexy'," Asbell notes.
In the Rod Stewart case, once the similarity was pointed out, Stewart "held [his] hand up straightaway" and acknowledged the melody wasn't entirely his. The dispute was settled out of court, with Stewart donating his royalties to UNICEF. And this is quite different from the Sheeran case — Stewart copied a distinctive melody, not just a chord progression.
Other cases have gone badly for artists. Katy Perry's "Dark Horse" was deemed to have infringed on "Joyful Noise," resulting in a $2.78 million judgment. The court found that using six "nearly identical" melodies was enough to warrant compensation, despite Perry claiming she'd never heard the original song.
So the line between inspiration and theft can be pretty thin. And with thousands of songs being released every day, similarities will happen by coincidence. But courts don't care about intent — they care about access and substantial similarity.
Detection Methods: From Expert Ears to AI Tools
Finding someone who stole your song used to be a simple process — you listened to both songs and yelled "that sounds like mine!" But the music industry needs something more solid than angry accusations.
For decades, music plagiarism has been spotted mainly by people with trained ears. "Traditionally, people's ears were used, supplemented by expert testimony. More recently, computer programs, or even AI-based technology, can be used to detect copying," explains Mark S. Lee, Partner at Rimon Law.
These experts, called forensic musicologists, break down songs into their basic parts and compare them side by side. They transcribe melodies, analyze chord progressions, and even examine the rhythm patterns to find matches.
And this approach gets more formal in court. Ben Michael, Attorney at Michael & Associates, points out how this plays out: "Proving plagiarism in music can actually be quite difficult. There are two key elements that have to be proven: access and substantial similarity. Access means that the person had access to the alleged copied song before creating their own song, which has to be proven beyond a reasonable doubt. Substantial similarity means proving that the elements of the song are similar enough to the original song to assume they could have been copied. This is a lot more difficult to prove, and typically musicologists are called upon to give their expert opinion."
Well, this works well enough, but the process has problems. For example, what sounds copied to one expert might sound like coincidence to another. And so each side brings their own music expert, leading to a battle of opinions rather than facts. These human-driven analyses can be quite subjective and influenced by all sorts of biases.
But technology is starting to change this. Spotify has been working on a "Plagiarism Risk Detector" that would scan new melodies against a database of existing songs. The system calculates a "similarity value" to show how close your tune is to something that already exists. Pretty handy for songwriters who want to check their work before they get sued.
AI might actually be better at this job than humans. Researchers at West Virginia University found that artificial intelligence can identify musical copyright infringement more accurately than human experts. The machines can analyze thousands of songs to understand what patterns are common in a genre versus truly unique elements that might have been copied.
For example, AI analysis could have helped in the Ed Sheeran case by showing that the four-chord sequence in question appears in hundreds of other songs, making it a common building block rather than stolen property.
But there's a problem with these new tools — false positives. Music is full of patterns that get reused all the time. An overeager algorithm might flag similarities that are just musical coincidences or genre conventions. A basic four-chord progression shows up in countless pop songs, but that doesn't mean they're all plagiarized versions of each other.
Read also: Sampling or Plagiarism? How to Sample Songs Legally And Creatively in Your Beats
The ideal detection system would need to understand context. It should know the difference between copying a distinctive melody (a potential copyright violation) and using common musical elements that nobody can own.
So far, the perfect balance between human ears and machine analysis hasn't been found. But as AI gets smarter, we might see plagiarism detection become more science than art. That could be good news for artists worried about accidental copying — and bad news for those hoping to sneak someone else's tune past listeners.
Lyrical Larceny: When Words Alone Are Stolen
You might think music plagiarism is all about stolen melodies and beats, but lyrics get ripped off too. And sometimes the words matter more than the music.
Song lyrics receive their own copyright protection, separate from the tune they're set to. "Copying of lyrics can be infringement because lyrics can be separately protectable under copyright law," explains Matthew Asbell, a partner at Lippes Mathias. "Imagine if poetry could be freely copied, but not when it was set to music. That would be an outrageous circumstance."
So yes, stealing someone's lyrics can land you in legal trouble, even if you pair them with completely different music. Although not all words are created equal when it comes to copyright.
Short phrases and common expressions don't qualify for protection. So lyrics need to show some originality to be protected. Mark S. Lee, Partner at Rimon Law, puts it simply: "Lyrics alone can be enough, so long as they are deemed original and more than generic short phrases."
But where's the line between common phrases and protected expressions?
The courts have been trying to figure that out. Taylor Swift faced a lawsuit over her hit "Shake It Off" because she used the phrase "players gonna play... and the haters gonna hate." Two songwriters claimed she stole it from their 2001 song "Playas Gon' Play." After five years of legal battling, the case was finally dropped in 2022. The court seemed to agree that these were common phrases in pop culture, not unique lyrical creations.
Asbell adds some nuance to how these cases are evaluated: "To be valid, a claim of infringement of lyrics would likely require that more than just a few words were copied, and that the defendant's use did not impart some completely new meaning or context to the words."
And when there's doubt, many artists choose the safer path of giving credit rather than fighting in court. Olivia Rodrigo added members of Paramore to the songwriting credits for her hit "good 4 u" when fans noticed similarities to Paramore's "Misery Business," as reported by Variety. This preemptive credit-sharing has become quite common in the industry.
And of course, the issue gets complicated because of AI. In October 2023, major music publishers sued AI company Anthropic for feeding copyrighted lyrics into its chatbot without permission. The publishers argued that no one can reproduce and distribute lyrics "to build its own business" without permission — and that includes AI systems.
So what should songwriters take from all this? Lyrics matter legally. Taking a unique verse or chorus from another song is asking for trouble. Using common phrases or generic expressions is generally fine. And that gray area in between? That's where lawyers make their money.
When Your Song Is Stolen: Legal Recourse and Protection
So you've written a song and you want to protect yourself from someone stealing it (or fight, in case it has already happened.) Here are some practical steps to protect your creation.
Step 1: Register Your Copyright ASAP
First things first — make sure your copyright is properly registered.
"Artists should register the copyrights in their work (and not rely on the outdated concept of sending themselves copies in postmarked mail)," advises Matthew Asbell, a partner at Lippes Mathias.
This costs between $45-$65 and gives you the legal standing to pursue infringement claims. Without registration, you can't file a lawsuit, and you won't be eligible for statutory damages of up to $150,000 per work.
Step 2: Document Everything
If you suspect copyright infringement, gather concrete evidence showing:
- Your original creation (dated recordings, drafts, session files)
- The alleged infringement (recordings, transcriptions)
- Side-by-side comparisons highlighting similarities
- Proof the other artist could have accessed your work
Step 3: Get Expert Opinions
Hire a musicologist to analyze both songs and determine if substantial similarities exist. Then consult with an intellectual property attorney to assess the legal strength of your case.
Asbell recommends: "Then, they should consult a lawyer about next steps, which may likely involve a cease-and-desist letter and possible litigation. If they have budgetary limitations, they can consider working with a lawyer on a contingency basis or obtaining pro bono legal services from a law firm, a law school clinic, or an organization like Volunteer Lawyers for the Arts. They should avoid making public statements about the possible infringement."
Step 4: Send a Cease and Desist Letter
Your attorney will draft a formal notice to the alleged infringer outlining:
- Your copyright ownership
- The specific elements being infringed
- Demands to stop using your work
- Request for compensation/credit
- Deadline for response
Many cases get resolved at this stage through negotiation.
Step 5: Consider DMCA Takedowns
If your song is being shared online without permission, you have another option. Mark S. Lee, Partner at Rimon Law, explains: "Also, what are called DMCA ('Digital Millennium Copyright Act') takedown notices can be considered if the infringed material was posted on a website by a third-party user."
These notices can get the infringing content removed from platforms like YouTube, Spotify, or social media sites.
Step 6: File a Lawsuit (If Necessary)
If negotiations fail, your attorney will file a copyright infringement lawsuit in federal court. Be prepared for:
- Filing fees (several hundred dollars)
- Legal expenses (potentially thousands)
- Discovery process (evidence gathering, depositions)
- Possible settlement negotiations
- Trial (if settlement isn't reached)
Many music cases settle before trial. Ed Sheeran settled multiple infringement claims rather than go to court.
Step 7: Know Your Potential Outcomes
If you win, you might receive:
- Monetary damages (actual losses or statutory damages)
- Court order to stop distribution of the infringing song
- Credit as songwriter
- Attorney's fees reimbursement
If you lose, you may be responsible for your legal costs and potentially the defendant's fees if your claim is deemed baseless.
With all that being said, having your copyright properly registered gives you options and leverage. Without it, your hands are pretty much tied if someone steals your music. And while litigation is always an option, many cases end with some form of compromise rather than total victory.
The AI Boom: New Complications in Copyright
AI has thrown a giant monkey wrench into the already complex machine of music copyright. The tech can now create pretty good songs in seconds and copy styles with scary accuracy. And the music industry surely has started fighting back immediately. As mentioned before, in late 2023, major music publishers sued AI company Anthropic for using song lyrics to train its chatbot without permission. And more recently, Sony Music, Universal Music Group, and Warner Records filed lawsuits against AI music generators Suno and Udio, claiming copyright infringement on an "almost unimaginable scale."
But who's actually responsible when an AI creates something that sounds like an existing song? Matthew Asbell, a partner at Lippes Mathias, explains the key question: "We don't really know the answer as yet regarding whether AI-generated music that was trained on copyrighted material infringes. The real question is whether the process of training the model on copyrighted material is itself an infringement of the copyright, as opposed to the AI-generated result which could possibly avoid being considered 'substantially similar' (it might even be possible to prompt the AI to avoid certain similarities."
The U.S. Copyright Office has made one thing clear, though: AI-generated works without meaningful human input can't be copyrighted and copyright protection requires human authorship. So if an AI writes a song by itself, that music effectively enters the public domain — it can be used by anyone.
This creates an interesting scenario that Asbell points out: "AI will definitely play a significant role once we learn from the courts whether it may be trained on copyrighted material. For one, because music generated by AI is often unprotectable under copyright law unless it has been further modified by a human, the proliferation of AI-generated music will likely greatly expand the public domain, making it easier to freely copy and use music. The problem is that people will likely misinterpret the law and not realize when they are overreaching in their copying until it is too late and they find themselves in court."
The legal system is still catching up. Mark S. Lee, Partner at Rimon Law, sums it up: "Courts are grappling with those issues at present, and we may have more clarity on the issue in a couple of years."
So what can we expect moving forward? 2025 will likely bring "more AI suits filed, both by individual creators and corporate rights holders, and courts will begin ruling on thorny AI issues such as training data and fair use.
Lee offers a broader perspective on copyright's evolution: "Since its inclusion in the Constitution, copyright law has always sought to address the tension between the understandable desire of creators to profit from their creations, and the understandable desire of the public to benefit from the free access to the works that have been created. That's why copyright law has always granted protection for what the Constitution calls 'limited times.' Compensating creators encourages the creation of more new works, while wider access to existing works enriches public discourse and enjoyment. How long that protection should last, and in what circumstances, is an ongoing public dialogue and debate.
"Thus, we see Congress enacting longer and longer terms of copyright protection, while technology makes it easier and easier to copy works that have been created. Legislative and judicial changes to copyright law at the end of the 20th century have facilitated the growth of entire industries that profit from copying others' copyrighted works without compensation, in circumstances copyright owners do not like and lobby to change. Copyright law will continue to change as technology continues to make copying easier, and Congress and the courts decide when and how copyright owners should be compensated, or not, for that copying."
We might also see new licensing models emerge — maybe collective licenses that let AI companies train their models on music catalogs for a fee, similar to how radio stations pay to play songs. And AI might eventually become a copyright enforcer itself, scanning the internet for plagiarized music.
For now, musicians should be careful about using AI tools. Training an AI on someone else's music or using AI-generated material could put you at legal risk. The rules are being written as we speak, and nobody wants to be the cautionary tale and become more famous because of a loud lawsuit (even if it changes the legal system and copyright protection for the better).
