By: Caitlin Gallagher ‘26
Volume XI – Issue I – Fall 2025
I. INTRODUCTION
Chances are, if you are involved in the Internet or if you are even a casual music enjoyer, it is almost certain that you have heard a song that includes sampling. Senior Lecturer at the Abbey Road Institute, Jason O’Bryan, defines sampling as “when you include an element of a pre-existing recording by someone else in your composition.” The sample can be anything that you’ve ‘sampled’ from another track; a rhythm, a melody, a beat, vocals or speech, which you then manipulate, edit, chop up, or loop to fit creatively within your work.” [1] Popular examples include Drake’s “Nice For What,” which sampled Lauryn Hill’s 1998 “Ex-Factor,” and Beyonce’s “Crazy in Love,” which sampled the Chi-Lites’ song released in 1970, titled “Are You My Woman (Tell Me So).” These songs are just two of hundreds of examples of hit songs that use the practice of sampling. Taking pieces of old songs is a way to introduce younger listeners to the hits of an older generation, but it also often provides a gateway for older listeners to connect with music being produced by younger generations.
Given the number of artists and producers who have taken to using this popular technique, it is hard to imagine the music industry as we know it today without the history of sampling, but new technological trends might reframe the way we understand what it means to sample and who or what is allowed to do so. We know some things about how courts think about originality as it relates to sampling and artists taking inspiration from one another, but how might that change in the age of artificial intelligence? Understanding the legal history of sampling in the music industry, coupled with analyzing the current legal landscape of artificial intelligence and copyright law, suggests that clear restraints will be placed on the use of artificial intelligence in sampling. Whether or not those restraints will be harsher than the ones that exist for human artists remains to be seen. To understand what shape those restraints might take, this paper will first seek to understand the history of the role of the legal system in music sampling, and then will turn to analysis for what implications that history, in addition to a growing body of legal scholarship on the topic, might reveal for how artificial intelligence will be treated when it comes to music copyright.
II. THE HISTORY OF SAMPLING AND THE LAW
In order to understand how the law might handle artificial intelligence sampling technology, it is important to first understand how sampling emerged and how the law treated it as it became more popular. The technique emerged in the 1940s after being pioneered by a French composer, Pierre Schaeffer, who developed it as a way to layer sounds over one another on tape recordings. Two decades later, sampling was made easier by a device called “The Mellotron.” The Mellotron is a keyboard that plays “different portions of magnetic tape as the keys were pressed, reproducing pre-recorded sounds that varied from strings, woodwinds, brass or even a choir.” [2] This device was a kind of proto-synthesizer similar to what might come to mind when you think of a modern-day DJ deck or studio mixing board. At the time, though, similar to AI, the new technology was revolutionary to the industry. In the 1970s, the process became digitized with the creation of synthesizers like the Synclavier and the Fairlight CMI—devices that were able to transform audio sources into digital files. [3] This is when the term “sampling” was first coined. Initially, the technology was expensive, hard to acquire, and difficult to use, but as it became more accessible, in terms of price and user-friendliness, sampling became more popular. Its popularity is evidenced by its pervasiveness in the Hip-Hop genre starting in the 1980s and 90s.
From the introduction of this technique, legal battles ensued over issues of ownership and originality. A famous example of this connection between the legal landscape and music sampling was the legal feud between The Turtles, an American rock group that reached the height of their popularity in the 1960s, and De La Soul, a New York hip-hop group that rose to fame in the late 1980s. In De La Soul’s song titled “Transmitting Live From Mars,” the group sampled a twelve-second portion of The Turtles’ 1969 song “You Showed Me.” According to a 1989 Los Angeles Times piece that covered the dispute, former Turtles members’ gripes with the sample were that they were not approached by De La Soul to get permission for the sample, and that they were not paid royalties for the song. In the lawsuit, it was reported that The Turtles were asking for more than two million dollars in payment. [4] The suit was settled outside of court, and thus the details are not publicly available, but the two groups reportedly settled the dispute for $1.7 million. [5]
Although the De La Soul case set no binding legal precedent because it never went to trial, it is one of many cases in the 1980s and 1990s that were an indication that the “wild west” of sampling was not made to last. In other words, the settlement indicated the desire for a limit of sorts on similarity between songs. One example of this is the case, Grand Upright Music v. Warner Bros. Records, Inc. (780 F. Supp. 182 (S.D.N.Y. 1991)). In this case, the first line of District Judge Kevin Duffy’s memorandum in the case was “Thou shalt not steal.” [6] This comment is made in reference to Biz Markie’s sampling of Raymond O’Sullivan’s song, “Along Again (Naturally).” [7] The case made clear that the “free-for-all” on sampling that may have existed before was certainly over. In order to comply with copyright law, artists looking to produce a song using sampling must obtain permission from the artist of the original work. [8] Because Biz Markie did not do that in this case, he was eventually ordered to pay restitution to O’Sullivan.
Case law, like the Bridgeport Music, Inc. v. Dimension Films (410 F.3d 792 (6th Cir. 2005)) case, has further solidified what private settlements imply: that there is a line artists can cross when it comes to sampling. Bridgeport deals in part with issues of licensing, but also addresses the issue of sampling. In this case, Dimension Films produced I Got The Hookup, a movie that included the song, “100 Miles And Running” by N.W.A. which sampled the Funkadelics song, “Get Off Your Ass And Jam.” [9] Dimension Films, the defendant in this case, argued that the sample was de minimis, an exemption encountered commonly in music copyright law that “permits the unauthorized sampling of sounds if the copying is trivial.” [10] Dimension Films also argued that the portion of “Get Off Your Ass And Jam” that was sampled was not itself original and therefore not covered under copyright law. To the point that the sample was not original, the Court disagreed. Dimension Films argued that the chords were too common for the sample to be considered original, but the Court explained that it was not the commonality of the chords, but rather “the use of and the aural effect produced by the way the notes in the chord are played” that mattered most in deciding whether the Funkadelics work was original enough to be protected by copyright laws. [11]
The U.S. District Court for the Middle District of Tennessee also disagreed with the defendant’s de minimis argument. The opinion explains that the burden of proof falls on the plaintiff in order to show how a sample does not qualify for a de minimis exemption. The plaintiff must show 1) proof of copying, and 2) that the copied work and the original work are substantially similar. [12] The Court explains in this opinion that one of the most common ways to ascertain whether two works are “substantially similar” is by assessing “whether an average lay observer would recognize the alleged copy as having been appropriated from the copyrighted work.” [13] In this case it was found that the first prong of the test was certainly satisfied (i.e. that “100 Miles And Running” certainly copied a section of “Get Off Your Ass and Jam” and looped it into their own track). However, the plaintiff’s argument still failed to satisfy the second prong, as the Court believed that no reasonable jury member, even if they were fans of “Get Off Your Ass And Jam” would recognize the sampled sections in “100 Miles And Running” to be something taken from the original, thus proving the lack of “substantial similarity” between the two tracks. [14]
Although this case did not rule that this particular sample was unlawful, like the Grand Upright Music v. Warner Bros. Records, Inc. case in 1991, this case also sets some ground rules when it comes to whether sampling is unlawful or not. In fact, this precedent is still being used to decide sampling issues in the courts in the modern day. A relatively recent example of this was the opinion in the Williams v. Gaye (No. 15-56880 (9th Cir. 2018)) case; a lawsuit filed by Marvin Gaye’s three children against Pharrell Williams and Robin Thicke over sampling in the 2013 song “Blurred Lines.” Gaye’s children alleged that “Blurred Lines” was an infringement on their copyright in their father’s hit 1977 “Got To Give It Up.” [15] Given their continuing relevance, it is important to understand how the precedents these cases set may affect the landscape of artificial intelligence music programs.
III. FAIR USE
Another important aspect of sampling and how it relates to newly introduced artificial intelligence is the concept of fair use. Like sampling cases, the concept of fair use is important to understand if we are to understand how music created by artificial intelligence might be treated in the upcoming years. Among the most comprehensive cases that govern fair use disputes in the music industry is Campbell v. Acuff-Rose Music Inc. (510 U.S. 569 (1994). The case has facts that are similar to the previous cases discussed. In 1989, the group 2 Live Crew released a song titled, “Pretty Woman,” which was written in a way that made it a quite obvious parody of the 1964 Roy Orbison song, “Oh, Pretty Woman.” The 2 Live Crew song sampled the iconic bass riff from the beginning of the Orbison song, and the subsequent lyrics were meant to comically satirize the original song and lyrics. [16] 2 Live Crew asked Acuff-Rose, a music publisher who owns the rights to the original Orbison song, for permission to use the song. Despite their request being denied, 2 Live Crew used the bass sample and satirical lyrics anyway. [17]
The Supreme Court rightfully pointed out in their opinion that, “From the infancy of copyright protection, some opportunity for fair use of copyrighted materials has been thought necessary to fulfill copyright's very purpose,” meaning that it promotes art and creativity. [18] Additionally, the Acuff-Rose case quotes the opinion from Emerson v. Davies (8 F. Cas. 615 (CCD Mass. 1845)) which states, "[i]n truth, in literature, in science and in art, there are, and can be, few, if any, things, which in an abstract sense, are strictly new and original throughout. Every book in literature, science and art, borrows, and must necessarily borrow, and use much which was well known and used before.” [19] The 1976 Copyright Act itself explains what fair use is, and indicates what is or is not included underneath the label: “the fair use of a copyrighted work, including such use by reproduction in copies or phonorecords or by any other means specified by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright.” [20] The act also gives courts tools to assess whether or not the use of copyright material is fair use. The Court must assess: “1) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes, 2) the nature of the copyrighted work, 3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and 4) the effect of the use upon the potential market for or value of the copyrighted work.” [21]
The Court found that the song was indeed a parody, which is covered by the fair use carved out in the statute, but the majority of their opinion discussed whether or not 2 Live Crew’s commercial distribution of their song affected whether or not their work was still protected as fair use. The Court found that, “the mere fact that a use is educational and not for profit does not insulate it from a finding of infringement, any more than the commercial character of a use bars a finding of fairness.” [22] Because of this, the Court in the Acuff-Rose case ruled that even though the 2 Live Crew song was being distributed for commercial purposes, as most songs are, the group did not violate the copyright of the original song because they had a fair use claim to the way they used the original to create their parodic version under the four guidelines the statute provides. Commercial use by itself is not enough to deny a fair use claim, parodies have historically been allowed by the Court, and the song had little potential for effect on the market for Orbison’s original song. Like in Bridgeport Music, Inc. v. Dimension Films, the Court did not rule that this specific sampling and inspiration-taking was not protected by the fair use clause, but it did offer guidance on what may or may not be acceptable. For example, there would likely never be a fair use exemption for a song that was merely copied from the original. When it comes to artificial intelligence in music, it is important to note how courts have treated fair use, because it is possible that fair use will govern how music created by artificial intelligence is treated.
IV. THE CURRENT LANDSCAPE: MUSIC AND ARTIFICIAL INTELLIGENCE
It is with this history in mind that we can approach the topic of artificial intelligence in the music industry. Like the synthesizers and keyboards that made sampling popular for the first time, artificial intelligence is the new technology that complicates existing notions of originality and copyright in the music industry. It should be unsurprising, then, that the industry is bracing for a new wave of litigation as a result of the rising popularity of this type of technology. The existing case law is fairly clear on how to treat sampled content of music that already exists, but what happens when generative artificial intelligence is producing music that imitates an artist’s style or voice, but was not actually created by that person themselves?
Questions of copyright ownership are more convoluted in cases like this, but luckily, there is some, albeit minimal, guidance on how cases involving artificial intelligence and music sampling might turn out. For example, Christopher Zirpoli, a Legislative Attorney for the Congressional Legal Service, has written a report relating to the topic for the Library of Congress’s Legal Sidebar. In his report, he primarily tackles how copyright laws more broadly might treat generative artificial intelligence. Zirpoli raises the interesting point that, traditionally, courts in the United States have held that in order for an artist to have access to the copyright they are entitled to under the Copyright Act, they must be a human author. [23] Put more specifically, Zirpoli explains that the U.S. Copyright Office has long declined to award copyright to works that are “produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.” [24] There is certainly a valid argument to be made that generative artificial intelligence programs are more than just mechanical processes that operate at random, as artificial intelligence programs are “technology that enable computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy.” [25] Artificial intelligence’s works fail the requirements for copyright protection might fall under the latter half of the Copyright Office’s requirement, as they do not necessarily require a creative, human input in order to be created. Because artificial intelligence is becoming increasingly competent at producing satisfactory outputs, and is designed to recreate human thinking by nature, human involvement is becoming less necessary. For example, the song, “How Was I Supposed to Know?” recently went viral on TikTok. The song was produced by an artist named, Xania Monet. As it turns out, there actually is no human artist named Xania Monet—she is an artificial intelligence avatar who created this song. She was created by a real person, an artist named Telisha Nikki Jones, but the song itself was not Jones singing or creating the melodies of Monet’s hit songs. Nevertheless, Jones recently fetched a $3 million record deal. [26]
All of that being said, even Zirpoli acknowledges that this rule leaves room for works that are mostly produced by artificial intelligence to gain copyright protections. He writes, “works containing AIgenerated material may be copyrighted under some circumstances, such as 'sufficiently creative' human arrangements or modifications of AI-generated material or works that combine AI-generated and humanauthored material.” [27] Even still, Zirpoli points to the fact that artificial intelligence guidance issued by the U.S. Copyright Office states that authors may claim copyright protection only “‘for their own contributions’ to such works, and they must identify and disclaim AI-generated parts of the works when applying to register their copyright.” [28]
If the guidelines and Zirpoli’s analysis are still unconvincing that works generated by artificial intelligence may not enjoy the same rights as works created by human authors, some cases suggest that works produced by artificial intelligence will not be awarded copyright. Zirpoli points to a recent case, Thaler v. Perlmutter (No. 23-5233 (D.C. Cir. 2025)). In that case, Thaler argued that a work produced by his artificial intelligence model should have access to copyright for the works it produced. The model was subsequently denied these rights by the U.S. Copyright Office, a decision that was upheld by the District Court for the District of Columbia. In the opinion, Judge Patricia Millett writes, “The Creativity Machine cannot be the recognized author of a copyrighted work because the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being.” [29] Keeping the Thaler case and Zirpolis’ helpful analysis in mind, it seems improbable that works generated by artificial intelligence will have the same access to copyright protections that human-authored works do anytime in the near future.
This relatively clear restriction is helpful in imagining a case where a music artist would have no copyright assertions over a piece of music created with artificial intelligence, and thus would likely have limited rights in the event that particular song was ever sampled by another artist. However, we are still left wondering what the outcome would be in a case where an artist’s likeness was used to create a “sample” of their work that never existed. One of the things that generative artificial intelligence programs have become eerily adept at is producing audio content that closely resembles human voices. There has been a rising phenomenon in the music industry where artificial intelligence has been used to mimic artists' voices, make audio recordings of that mimicry, and then use those audio recordings as samples in their own new songs.
One well-known example of this occurred during the infamous social media squabble between prominent rap artists Drake and Kendrick Lamar. In his diss-track against Lamar, titled “Taylor Made Freestyle,” Drake sampled an artificial intelligence-produced recording of Snoop Dogg’s voice in addition to Tupac Shakur’s voice, an artist who has been deceased since 1996. [30] The song, while not officially released to streaming platforms, was posted by Drake to his own social media and still exists in an unofficial capacity on online platforms like SoundCloud and Genius Lyrics. [31] In the song, artificially generated Tupac raps the introduction and the first verse, artificially generated Snoop Dogg raps the second verse, and Drake comes in only to conclude the song. [32] Shakur’s estate was reportedly infuriated by the release of the track, and threatened legal action against the Canadian rap star. [33]
In cases like these, we are still left with the question of whether or not these artificial samples involve fair use. There is enough evidence to suggest that the works would not be treated as copyright protected “originals,” which would make them eligible to be sampled under fair use just as Roy Orbison’s music was sampled by 2 Live Crew in 1989. If they are to be considered works that the original artist has copyrights to, it is entirely possible that songs that sampled them would be within their fair use rights as explained in the Campbell v. Acuff-Rose Music Inc. case. However, if there is no copyright to be had for these works, like the Thaler v. Perlmutter case and Christopher Zirpoli’s research suggest there will not be, the issue of whether or not these artificially generated “samples” would be protected under fair use is a kind of legal limbo. In the case of the latter, perhaps they will fall more generally under case law that has generally governed sampling as explained by the Grand Upright Music v. Warner Bros. Records, Inc. case and the Bridgeport Music, Inc. v. Dimension Films case. Grand Upright Music explains that artists must give permission for their originals to be sampled, which seems relatively straightforward, but it still remains unclear whether or not artificially generated works are to be considered originals of the artists they imitate. Perhaps a rule created in the image of the Grand Upright Music rule will require artists using artificially generated samples to ask permission of the artist the artificial intelligence is copying in order to be considered legally valid. Keeping in line with Bridgeport Films, unauthorized replications of an artist’s voice or music style would only be exempt from legal action if they were de minimis samples.
V. CONCLUSION
The issue of sampling has been an issue in the realm of music and the law for nearly a century, and with the rise of artificial intelligence, a new wrinkle has been introduced into the legal landscape. Although many legal questions have so far gone unanswered, there is a sturdy legal framework that has existed for many years as a result of cases about sampling. There can be no understanding about the shape these new restraints on artificial intelligence and sampling will take without understanding the precedents of sampling and music copyright law. Unfortunately for the project of determining concrete case law for issues such as these, there has yet to be a lawsuit that conclusively answers these questions. Many were quick to denounce the practice of artificially generated music entirely, but an understanding of the technological history of music provides a pathway for a more nuanced perspective. If one thinks about how new music technology has been received, even the sampling technology discussed at the beginning of this paper, it has often not been positive. Some saw the introduction of Synclavier and the Fairlight CMI as tools to sample older music as the death of creativity and originality among the younger generations. Looking back at this history now, it would be an understatement to say that those opinions were incorrect. The process of sampling has completely transformed the music industry. It is hard to imagine a modern music landscape absent of the amazing works that include sampled portions.
There is good reason to suspect that artificial intelligence has the same capacity to change the music industry as well. As it was with sampling originally, the industry finds itself in an odd limbo when it comes to sampling and artificial intelligence at the moment. However, if history is useful in determining the future—which in the field of law it so often is—then the use of artificial intelligence will be limited in the next couple of years. The precedent set by cases like Grand Upright Music v. Warner Bros. Records, Inc., Bridgeport Music, Inc. v. Dimension Films, Thaler v. Perlmutter, and Campbell v. Acuff-Rose Music Inc. gives some insight into how these artificial intelligence may be treated. There is ample evidence to suggest that those who utilize artificial intelligence models will have limited copyrights to works that are machinecreated, but limiting its use entirely is not the best solution. There is still opportunity for artificial intelligence to create unique and interesting new music that can be governed by frameworks that have already been created. Answering the questions the introduction of artificial intelligence in the music industry has posed is complicated, but it is not impossible. By turning to the history of sampling and fair use in music law, some answers begin to emerge.
Endnotes
[1] Abbey Road Institute, Sampling: Its Role In Hip Hop and Its Legacy In Music Production, ABBEY ROAD INSTITUTE (October 19th, 2020), https://abbeyroadinstitute.co.uk/blog/sampling-role-in-hip-hop-and-its-legacy-inmusic-production/.
[2] Abbey Road Institute, Sampling: Its Role In Hip Hop and Its Legacy In Music Production.
[3] Abbey Road Institute, Sampling: Its Role In Hip Hop and Its Legacy In Music Production.
[4] L.A. Times Archives, “Sampling Wars,”, Los Angeles Times (July 23rd, 1989), https://www.latimes.com/archives/la-xpm-1989-07-23-ca-392-story.html.
[5] The Economic Times, “Fights for Rights: Glimpses of Music Copyright Disputes from the Past,” The Economic Times, March 27, 2017, https://economictimes.indiatimes.com/entertainment/fights-for-rights-glimpses-of-musiccopyright-disputes-from-the-past/sp-balasubrahmanyam-vs-ilaiyaraaja/slideshow/57850544.cms.
[6] Grand Upright Music v. Warner Bros. Records Inc., 780 F. Supp. 182 (S.D.N.Y. 1991).
[7] Grand Upright Music v. Warner Bros. Records Inc., 780 F. Supp. 182.
[8] Grand Upright Music v. Warner Bros. Records Inc., 780 F. Supp. 182.
[9] Bridgeport Music, Inc. v. Dimension Films, 230 F. Supp. 2d 830 (M.D. Tenn. 2002)
[10] Kyle Attisha, “The De Minimis Exception: The Need For A Copyright Office Study Regarding Music Sampling,” Dartmouth Law Journal 21 no. 2 (Fall 2023): 36-63, https://dartmouthlawjournal.org/article/129816.
[11] Bridgeport Music, Inc. v. Dimension Films, 230 F. Supp. 2d 830 (M.D. Tenn. 2002).
[12] Bridgeport Music, Inc. v. Dimension Films, 230 F. Supp. 2d 830.
[13] Bridgeport Music, Inc. v. Dimension Films, 230 F. Supp. 2d 830.
[14] Bridgeport Music, Inc. v. Dimension Films, 230 F. Supp. 2d 830.
[15] Williams v. Gaye, No. 15-56880 (9th Cir. 2018).
[16] Campbell v. Acuff-Rose Music (92-1292), 510 U.S. 569 (1994).
[17] Campbell v. Acuff-Rose Music (92-1292), 510 U.S. 569 (1994).
[18] Campbell v. Acuff-Rose Music 510 U.S. 569.
[19] Campbell v. Acuff-Rose Music 510 U.S. 569.
[20] 17 U.S.C. § 107 (1976).
[21] 17 U.S.C. § 107 (1976).
[22] Campbell v. Acuff-Rose Music 510 U.S. 569.
[23] Christopher T. Zirpoli, Generative Artificial Intelligence and Copyright Law, CONGRESS.GOV (July 18th, 2025),https://www.congress.gov/crsproduct/LSB10922#:~:text=Granting%20that%20human%20authors%20may,applying%20to%20register%20their% 20copyright.
[24] Zirpoli, Generative Artificial Intelligence and Copyright Law.
[25] Cole Stryker & Eda Kavlakoglu, What is artificial intelligence (AI)?, IBM (Last visited December 3rd, 2025), https://www.ibm.com/think/topics/artificial-intelligence.
[26] Ben Naddaff-Hafrey, The irony in an AI musician singing “How Was I Supposed to Know?”, MS NOW (November 10th, 2025), https://www.ms.now/opinion/msnbc-opinion/xania-monet-ai-song-telisha-jones-billboardhit-rcna242478.
[27] Zirpoli, Generative Artificial Intelligence and Copyright Law.
[28] Zirpoli, Generative Artificial Intelligence and Copyright Law.
[29] Thaler v. Perlmutter, No. 23-5233 (D.C. Cir. Mar. 18, 2025).
[30] Edward Lee, “AI and the Sound of Music,” Yale Law Journal 134 (November 22, 2024): 187–236, https://doi.org/https://yalelawjournal.org/essay/ai-and-the-sound-of-music.
[31] Genius Lyrics, Taylor Made Freestyle Lyrics, GENIUS.COM (Last visited December 3rd, 2025), https://genius.com/31498244.
[32] Genius Lyrics, Taylor Made Freestyle Lyrics.
[33] Lee, “AI and the Sound of Music.”