Tift Merritt’s most popular song on Spotify, “Traveling Alone,” opens in a new tab, is a ballad with lyrics about solitude and freedom
.
Prompted by Reuters to create “an Americana song in the style of Tift Merritt,” the artificial intelligence music website Udio instantaneously developed “Holy Grounds,” an open new tab ballad with lyrics about “driving old backroads” while “watching the fields and skies shift and sway.”
Merritt, a Grammy-nominated singer and songwriter, told Reuters that Udio’s “imitation” “doesn’t make the cut for any album of mine.”
“This is a great demonstration of the extent to which this technology is not transformative at all,” according to Merritt. “It’s stealing.”
Merritt, a veteran artist rights campaigner, isn’t the only musician raising concerns. In April, she signed an open letter with Billie Eilish, Nicki Minaj, Stevie Wonder, and dozens of other artists warning that AI-generated music based on their recordings might “sabotage creativity” and push human artists out of the spotlight.
The major record labels are also apprehensive. Sony Music (6758.T), Universal Music Group, and Warner Music sued Udio and another music AI business named Suno in June, ushering the music industry into high-stakes copyright conflicts over AI-generated content that are only now beginning to play out in court.
“Ingesting massive amounts of creative labor to imitate it is not creative,” said Merritt, an independent musician whose original record label is now owned by UMG but who claims she has no financial stake in the corporation. “That’s stealing in order to be competition and replace us.”
When contacted to comment on this topic, Suno and Udio referred to previous public statements supporting their technology. They submitted their initial reply in court on Thursday, denying any copyright infringement and claiming that the actions were intended to hinder smaller competitors. They likened the labels’ criticisms to previous industry fears over synthesizers, drum machines, and other advances that replaced human musicians.
Uncharted ground
The firms, which have both received venture capital backing, have stated that they prohibit users from writing songs that directly imitate top musicians. However, the new claims allege that Suno and Udio may be directed to duplicate sections of songs by Mariah Carey, James Brown, and others, as well as to impersonate the voices of musicians such as ABBA and Bruce Springsteen, demonstrating that they exploited the labels’ database of copyrighted recordings to train their systems.
According to Mitch Glazier, CEO of the Recording Industry Association of America (RIAA), the lawsuits “document shameless copying of troves of recordings in order to flood the market with cheap imitations and drain away listens and income from real human artists and songwriters.”
“AI has great promise – but only if it’s built on a sound, responsible, licensed footing,” Glazier told reporters.
When asked to comment on the lawsuits, Warner Music directed Reuters to the RIAA. Sony and UMG have not responded.
The labels’ arguments resemble those made by writers, news sites, music publishers, and others in high-profile copyright cases against chatbots such as OpenAI’s ChatGPT and Anthropic’s Claude, which utilize generative AI to create text. These lawsuits are still pending and in the early stages.
Both sets of instances raise significant legal problems, such as whether the law should make an exception for AI’s use of copyrighted content to develop something new. The record labels’ cases, which could take years to resolve, also present issues specific to their subject matter – music.
The interplay of melody, harmony, rhythm, and other characteristics can make it more difficult to establish when parts of a copyrighted song have been infringed than works such as written text, according to Brian McBrearty, a musicologist who specializes in copyright analysis.
“Music has more factors than just the stream of words,” McBrearty added. “It contains pitch, rhythm, and a harmonic context. It’s a more diverse blend of elements that make it a little less straightforward.”
Some allegations in AI copyright lawsuits may be based on comparisons between an AI system’s output and the material purportedly used to train it, necessitating the type of examination that judges and juries have faced in music cases.
In a 2018 ruling that a dissenting judge labeled “a dangerous precedent,” Robin Thicke and Pharrell Williams lost a dispute brought by Marvin Gaye’s estate over the likeness of their song “Blurred Lines” to Gaye’s “Got to Give It Up.” However, singers such as Katy Perry and Ed Sheeran have responded to similar complaints about their own songs.
Suno and Udio contended in very identical court filings that their outputs do not violate copyrights and that U.S. copyright law protects sound recordings that “imitate or simulate” previously recorded music.
“Music copyright has always been a messy universe,” said Julie Albert, an intellectual property lawyer at New York-based legal firm Baker Botts who is following the new cases. Even without that difficulty, Albert claims that rapidly expanding AI technology is causing fresh uncertainties at all levels of copyright law.
Who is entitled to fair use?
The complexities of music may be less important in the end if, as many believe, the AI cases reduce down to a “fair use” defense against infringement accusations – another area of US copyright law fraught with unknowns.
Fair use encourages freedom of speech by permitting illegal use of copyrighted works under specific conditions, with courts frequently focused on whether the new use alters the original works.
Defendants in AI copyright disputes have claimed that their products make fair use of human creations, and that any court finding to the contrary would be terrible for the potentially multi-trillion-dollar AI industry.
Suno and Udio stated in their responses to the labels’ complaints on Thursday that using existing recordings to help people produce new music “is a quintessential ‘fair use.'”
Legal experts say fair use could make or break the cases, but no court has decided on the question in the context of artificial intelligence.
Albert believes that music-generating AI companies will have a more difficult time showing fair use than chatbot developers, who can summarize and synthesize text in ways that courts are more likely to consider transformative.
Consider a student using AI to create a report about the US Civil War that includes material from a novel on the subject, as opposed to someone asking AI to create new music based on existing music, she added.
Student example “certainly feels like a different purpose than logging onto a music-generating tool and saying ‘hey, I’d like to make a song that sounds like a top 10 artist,‘” stated Albert. “The purpose is pretty similar to what the artist would have had in the first place.”
A Supreme Court decision on fair use last year could have a significant influence on music cases because it focused primarily on whether a new use serves the same commercial purpose as the original work. This claim is central to the Suno and Udio complaints, which allege that the companies utilize the labels’ music “for the ultimate purpose of poaching the listeners, fans, and potential licensees of the sound recordings [they] copied.”
Merritt expressed concern that technology companies may utilize AI to replace artists like her. She explained that if musicians’ songs can be extracted for free and used to duplicate them, the economics are simple.
“Robots and AI do not get royalties,” she stated.
