By Chris Cooke & Investigative Desk
Published: Industry Analysis & Legal Report
Special Focus: Universal Music Group, Warner Music Group, American Federation of Musicians (AFM), and Generative AI Law
Executive Overview
As the global music industry rushes headlong into a gold rush of high-profile artificial intelligence licensing deals, a foundational legal battle is brewing behind closed doors. At the center of the storm is a high-stakes lawsuit filed by the American Federation of Musicians (AFM)—one of the preeminent musician unions in the United States—against two industry behemoths: Universal Music Group (UMG) and Warner Music Group (WMG).
Initially filed in June and progressing through contentious new court filings, the lawsuit strikes at the very heart of how recorded music is monetized, licensed, and protected in the age of generative machine learning. The core dispute centers on whether major record labels have a contractual obligation to compensate session musicians and background performers when catalog recordings are licensed to train generative AI models, such as those developed by tech platforms Udio and Suno.
The legal theater hinges on the interpretation of a single, highly contested provision within the industry-standard Sound Recording Labor Agreement (SRLA): the "new use" clause. The union argues that feeding copyrighted master recordings into AI models constitutes an unprecedented, uncompensated exploitation of musicians’ work, triggering mandatory payout structures. Conversely, the major labels are moving aggressively for a full dismissal, arguing that the clause is unambiguous, legally inapplicable to technologies that lack a pre-existing corresponding union agreement, and fundamentally misread by the labor organization.
This conflict is more than a localized contract dispute; it is a bellwether for the entire creative economy. As record companies increasingly partner with tech startups to secure training data, session musicians—the invisible architects of modern recorded music—risk being left entirely out of the financial equation. If the labels succeed in dismissing the case, it could establish a legal precedent that greenlights the uncompensated harvesting of human artistry for machine learning. If the AFM prevails, it could force a massive financial reckoning across the music business, setting a blueprint for how AI companies and major rightsholders must distribute licensing revenues down to the creator level.
Detailed Chronology: From AI Gold Rush to Federal Courtroom
1. The Proliferation of AI Licensing Deals
Over the past twenty-four months, generative AI has transformed from a fringe technological curiosity into the music industry’s most polarizing frontier. Major music conglomerates have pivoted swiftly from viewing AI strictly as an existential copyright threat to treating it as a lucrative licensing market. Universal Music, Warner Music, and Sony Music have entered into strategic partnerships and legal settlements with various AI startups, platform developers, and tech giants.
These deals typically allow AI developers to train their neural networks on vast corpuses of recorded music, promising to generate new compositions, vocals, and instrumental tracks that mimic human styles. However, while executives at UMG and WMG negotiate multi-million-dollar agreements, a critical constituency remains entirely sidelined: the session musicians and background performers whose performances constitute the training data.
2. June 2024: The AFM Files Suit
Recognizing the existential threat posed to its membership base—whose performances are effectively being reverse-engineered by AI models to compete directly against them—the American Federation of Musicians took legal action. In June, the union filed a federal lawsuit against both Universal Music and Warner Music.
The complaint alleged a systematic breach of contract. Specifically, the AFM pointed to the labels’ recent licensing agreements with generative AI platforms like Udio (involve both UMG and WMG) and Suno (involving WMG). According to the union, these deals represent a textbook violation of the "new use" clauses embedded within their collective bargaining frameworks.
3. Fall 2024: The Motion to Dismiss and New Filings
As the case advanced into the autumn months, the legal sparring intensified. Universal Music and Warner Music filed formal motions seeking to throw out the lawsuit in its entirety. The majors argued that the plaintiff’s legal theories were legally untenable, rooted in a misinterpretation of contract law, and fundamentally flawed.
Responding in kind, both the union and the major labels submitted fresh rounds of briefs to the court. These filings crystallized the central philosophical and legal chasm dividing labor and management: Can a decades-old collective bargaining agreement crafted for physical media and broadcast innovations be stretched to govern cutting-edge machine learning? Or, conversely, are the labels attempting to exploit a regulatory vacuum to evade paying the creators who built their master catalogs?
Supporting Context & Metrics: Decoding the "New Use" Clause
To understand the legal mechanics of the AFM’s lawsuit, one must examine the Sound Recording Labor Agreement (SRLA). The SRLA is the foundational collective bargaining agreement that governs the employment, compensation, and utilization of session musicians working on commercial sound recordings in the United States.
The Anatomy of a "New Use"
Within the SRLA sits a vital protective mechanism known as the new use clause. Historically, this clause was designed to protect musicians when a sound recording produced for one specific medium was subsequently repurposed for another, distinct medium that was not anticipated or monetarily accounted for when the recording session originally took place.
For instance, if a track recorded strictly for a vinyl album was later licensed for a television commercial, a feature film, or a nascent digital format, the record label was contractually bound to:
- Notify the union of the new utilization; and
- Compensate the musicians who performed on the original recording as if they had been brought back into the studio to record specifically for that new purpose, utilizing standard session rate cards.
The Core Conflict: AI as a "New Use"
The AFM’s argument rests on a straightforward logical syllogism:
- Premise 1: The SRLA covers recordings made by union musicians.
- Premise 2: Licensing those recordings to train a generative AI model is unquestionably a new, distinct purpose not originally contemplated when the tracks were cut.
- Conclusion: Therefore, the new use clause is triggered, requiring the labels to inform the union and pay out compensatory damages based on objective economic metrics within the agreement.
However, the major labels have constructed a sophisticated structural defense to neutralize this argument.
Official Statements and Legal Posturing
The briefs submitted by Universal Music, Warner Music, and the AFM reveal radically divergent views on contract interpretation, statutory intent, and linguistic clarity.
The Major Labels’ Defense: "Clear, Unambiguous, and Inapplicable"
In its latest court filings, Universal Music insists that the disputed SRLA clause is entirely "clear and unambiguous," leaving no room for the expansive, flexible interpretation advanced by the union. UMG characterizes the AFM’s legal strategy not as a serious contractual argument, but rather as "a scattershot series of arguments that obscure rather than confront what the contract makes clear."
The majors’ primary legal defense hinges on the conditional mechanics of how the new use rate is calculated. The clause dictates that when a label uses a covered recording for a purpose not covered by the agreement, it must pay:
“…an amount equal to all payments that would be required under the AFM agreement that would then be effective if the recording were originally made for the purpose set forth under that agreement.”
According to UMG and WMG, this language requires the existence of a separate, specific AFM agreement governing the exact new purpose in question in order to import a valid rate. Because no collective bargaining agreement or standard union rate card currently exists for generative AI training sessions, the majors argue that the mechanism breaks down completely.
In short, the labels maintain that the new use clause "does not state a rate" of its own; instead, it "imports one from the separate AFM agreement governing the new purpose." Because the union itself has conceded—including during pre-motion meet-and-confer sessions—that no such AI-specific agreement exists, the majors argue that the clause cannot legally be invoked. Consequently, they contend the judge must dismiss the lawsuit.
The Union’s Rebuttal: Textual Mandates and Objective Metrics
The AFM firmly rejects the labels’ restrictive reading, offering a very different interpretation of the contract’s structural integrity and intent.
In its counter-filings, the union insists that the absence of a pre-existing AI agreement does not nullify the label’s underlying payment obligations. The text of the SRLA, the union argues, "imposes a mandatory payment obligation when a signatory company puts a covered recording to a purpose not covered by the SRLA."
Addressing the absence of a specific AI rate card, the AFM contends that courts do not need a bespoke, purpose-built AI agreement to calculate financial harm. Existing provisions within the broader SRLA—including baseline session rates, streaming compensation metrics, and digital sampling fees—provide sufficient, objective economic benchmarks from which judicial damages can be accurately calculated.
Furthermore, the AFM has adopted a tactically advantageous posture regarding the legal standard of contract ambiguity. While the major labels desperately argue that the clause is unambiguous and favors their dismissal bid, the union freely concedes that the provision is "reasonably susceptible to more than one interpretation."
Under civil procedure rules, if a plaintiff can convince a federal judge that a contractual clause is ambiguous and that the union’s interpretation is legally "plausible," the case must survive the motion to dismiss and proceed to the discovery phase. The AFM is fully confident that its reading clears this judicial threshold.
Broader Industry Implications: Consent, Compensation, and Copyright
While the immediate battle is being fought over the fine print of the SRLA in a federal courtroom, the ripple effects of this lawsuit extend far beyond the direct participants. The broader music ecosystem is currently grappling with two deeply contentious questions regarding generative AI:
- The Consent Debate: Will major rightsholders seek explicit, granular consent from individual featured artists, session players, and songwriters before opting master catalogs into AI training pipelines?
- The Revenue-Sharing Debate: How will any windfalls generated by corporate AI licensing deals be equitably distributed down the chain to the actual creators who breathed life into the recordings?
The Plight of the Session Musician
At present, commercial realities lean heavily toward corporate consolidation. While top-tier pop stars and legacy heritage acts negotiate bespoke opt-out clauses or lucrative equity stakes in AI ventures, session musicians—the drummers, guitarists, brass players, and string sections who perform anonymously on thousands of commercial tracks—are routinely ignored.
Nobody in corporate boardrooms is currently talking about consulting or cutting royalty checks to session players, despite the undeniable reality that modern AI models learn musical phrasing, timing, texture, and harmony directly from those performers’ recorded output. Once trained, these AI models actively compete against human musicians for commercial synchronization, background scoring, and commercial production work.
Consequently, any meaningful financial redress for session musicians in the AI era will likely depend on a two-pronged strategy: aggressive union litigation (such as the AFM’s current lawsuit) paired with sweeping statutory copyright law reform at the federal level.
Future Outlook: What Lies Ahead for the AFM v. Major Labels Suit
As the presiding federal judge weighs the cross-motions to dismiss, the stakes for both labor and management could not be higher.
- If the Labels Win: A dismissal would validate the majors’ narrow reading of legacy labor agreements, effectively insulating record labels from having to share AI licensing revenues with the session musicians whose work trained the models. It would send a chilling message to creative labor unions, proving that decades-old collective bargaining frameworks are structurally unequipped to handle technological paradigm shifts without explicit legislative updates.
- If the Union Wins: Surviving the motion to dismiss will force Universal Music and Warner Music into the grueling, expensive discovery phase of litigation. It would compel the labels to turn over internal communications, financial details of their deals with Udio and Suno, and strategic roadmaps regarding AI data acquisition. More importantly, it would create immediate leverage for the AFM to force industry-wide negotiations, establishing a binding precedent that AI training constitutes a compensable "new use" under American labor law.
For now, the legal briefs have been filed, the arguments have been heard in chambers, and the music industry waits anxiously for a judicial ruling that could redefine the economic architecture of recorded music for generations to come.
