In the grand halls of the United Nations in Geneva, the question posed by ABBA legend and CISAC president Björn Ulvaeus was as hauntingly simple as the melodies that defined a generation. Standing before a global audience at the "AI for Good" Global Summit on July 13, 2026, the music icon asked a pointed, rhetorical question: "Good for whom?"
This query serves as the epicenter of a widening rift between the Silicon Valley architects of generative artificial intelligence and the global creative community. As AI models ingest the sum total of human cultural output to learn the nuances of harmony, rhythm, and lyrical structure, the artists who provided the "raw material" are finding themselves sidelined. For Ulvaeus, the issue is not a Luddite-style rejection of technology, but a fundamental demand for agency, compensation, and a seat at the table.
A Chronology of Conflict: From Innovation to Indignation
The current tension is the culmination of years of rapid, often unchecked, technological acceleration.
- The Early 2020s: The rise of generative AI began as a novelty, with experimental models producing rudimentary audio clips.
- The "Voyage" Precedent: In 2022, ABBA set a high-water mark for "ethical AI" with their Voyage residency in London. By utilizing motion capture and machine learning to create digital avatars, the band proved that AI could enhance, rather than replace, human performance—provided there is "consent, participation, and compensation."
- The 2025 Inflection Point: By mid-2025, the music industry faced a stark reality: major labels and publishers were quietly negotiating blanket licensing deals with AI companies, often without the explicit, individual consent of the songwriters and session musicians whose work fed the algorithms.
- June 2026: The American Federation of Musicians (AFM) escalated the conflict by filing a landmark lawsuit against Universal Music Group (UMG) and Warner Music Group (WMG), alleging that these corporate titans were monetizing artists’ work through AI licensing deals while denying the creators their fair share of the proceeds.
- July 2026: Björn Ulvaeus’s keynote address at the UN, where he formalized the argument that the industry must treat AI developers as business partners who owe a debt to the creators whose works constitute their training data.
Supporting Data: The Anatomy of the Training Dispute
The core of the legal and moral struggle lies in the technical nature of "training." As Ulvaeus pointed out during his speech, there is a common misconception that AI "samples" songs in the traditional, copyright-infringing sense. Instead, these large language and audio models "learn relationships across billions of examples."
The data is clear: without the massive influx of copyrighted recordings, sheet music, and lyrical databases, these AI tools would be functionally inert. They rely on the "raw material" of human experience—grief, love, and cultural identity—to generate synthetic output.
Critics of the current AI model, including the 31 music rights organizations that penned a joint letter in mid-2026, argue that this is a case of systemic hypocrisy. Labels and publishers often demand that AI companies pay for the rights to their catalogs, yet they frequently fail to flow that revenue down to the individual songwriters and performers. As one industry insider noted, the "infrastructure for distribution" already exists—modeled after the collective licensing systems that have governed radio and streaming for decades—but the political and corporate will to implement it remains largely absent.
Official Responses and the Corporate Stance
The response from the music industry and AI developers has been varied and often defensive.
Universal Music Group and Warner Music Group have maintained that their licensing agreements are necessary to protect the value of their catalogs and provide a framework for legal AI development. However, the AFM’s lawsuit suggests a deep-seated frustration: the union argues that defendants have "refused to provide information to the AFM about which recordings and whose work is being licensed," creating an opaque environment that leaves artists in the dark about how their life’s work is being deployed.
At the UN summit, other industry leaders attempted to bridge the gap. John Legend, speaking alongside executives from NVIDIA, Splice, and Stability AI, echoed a sentiment of cautious optimism. "I want us to treasure the role that great art plays in our lives," Legend stated. "Treasuring also means requiring policy which protects the creators, so it is a viable career for artists."
Yet, the tension remains palpable. When lawyers argue that "blanket licensing" in standard record contracts provides labels with the right to train AI on an artist’s work without further approval, they are essentially stripping the artist of the power of consent. As attorney Jason Boyarski noted, some labels feel they "technically don’t need special approvals," a stance that threatens to turn the relationship between creators and labels from one of collaboration to one of litigation.
Implications for the Creative Economy
The implications of this standoff are profound, reaching far beyond the balance sheets of record labels. If the current trend of "free extraction" continues, the incentive structure for future generations of artists may collapse.
1. The Death of the "Professional Creator"
If a songwriter cannot earn a living because their work is being used to train a model that effectively generates competition for them, the barrier to entry for the arts becomes insurmountable. As Ulvaeus warned, if the well of human creativity runs dry because it is no longer a viable career path, AI will eventually run out of the very "original, new human creativity" it requires to evolve.
2. The Legal Precedent
The lawsuits currently winding through the courts will establish whether the digital age will prioritize the "right to train" or the "right of ownership." A victory for the artists would likely mandate a collective licensing model similar to the one proposed by Ulvaeus, where a percentage of AI subscription revenue is redirected to the creators of the training data.
3. A New Definition of "Fair Use"
The debate forces a re-evaluation of intellectual property. Is a "new synthesis" created by an AI truly original, or is it a derivative work of the billions of data points that informed its output? By moving the focus away from "tracing outputs" and toward "compensating inputs," Ulvaeus has provided a potential roadmap for a compromise that respects both the technology and the artist.
Conclusion: A Fork in the Road
As the dust settles on the AI for Good summit, the music industry finds itself at a precarious crossroads. The path forward requires more than just technological prowess; it requires a renewed social contract.
Björn Ulvaeus’s vision—a seat at the table for the creators—is not merely a plea for charity. It is a pragmatic warning. The history of music technology, from the advent of the phonograph to the disruption of Napster and the rise of streaming, shows that industries only survive when they find a way to honor the creator.
"Human creativity is not the enemy of artificial intelligence," Ulvaeus reminded the summit. "It is the reason artificial intelligence exists." Whether the next decade is defined by a collaborative partnership between human artistry and machine logic, or by a fractured, litigious landscape of "free extraction," will depend on the decisions made in these boardrooms and courtrooms today.
As the world watches, one thing is certain: the music will continue to play, but its future depends entirely on whether those who compose it are recognized as the partners they are, or the obstacles they are currently being treated as. The promise of 1791, when France first recognized the author’s right to their work, is being tested in the digital arena. It is up to this generation of policymakers and tech giants to ensure that the promise remains intact for the next century of creators.
