In 1927, The Jazz Singer became one of the first films with synchronized sound. Hollywood studios argued at the time that the technology was so disruptive it exempted them from paying royalties to the creators whose works they used. Nearly a hundred years later, at the general assembly where CISAC celebrated its centenary in Paris, director general Gadi Oron observed that artificial intelligence companies are making the same argument. They claim their work is transformative, that training on protected works does not require payment, and that generating value from human creativity without sharing that value with creators is a defensible position.
That analogy frames everything else that happened that day. The conflict between CMOs and technology platforms is not structurally new. The actor changes; the mechanism repeats. A new technology presents itself as a different category, one to which existing rules supposedly do not apply.
The case that could set the standard
CISAC president Björn Ulvaeus pointed to a specific lawsuit with a firm date on the calendar. The dispute is between GEMA and Suno, being heard in Munich. According to Ulvaeus, the outcome of that case functions as a hinge point for the entire music industry. A Suno win would undermine the legal basis for licensing deals covering AI-generated music. A Suno loss would consolidate licensing as the de facto standard for the sector.
That makes it a useful reference point for any CMO evaluating how to position itself against generative AI platforms, because the ruling will establish precedent well beyond Germany.
The regulatory front in France
Dean Ormston, chair of the CISAC board, called for public support behind the Darcos Bill, a proposed French law that would introduce a presumption of creative content use by AI platforms. In practice, this inverts the burden of proof. Rather than rights holders having to demonstrate their work was used without authorization, platforms would have to prove they did not use it.
This is a meaningfully different regulatory model from the one that prevails today, where the difficulty of accessing training data typically puts creators at a disadvantage when seeking recourse.
The impact already being felt
The most operational perspective came from composer Simon Franglen. He warned that generative AI will eliminate the bottom third of his business within two to three years, specifically the segment covering jingles, commercial music, and children's TV content. That segment has historically served as a training ground and first income source for young composers.
Franglen also raised a discoverability problem. If the catalog available on streaming platforms becomes saturated with AI-generated music, it becomes harder for listeners to find relevant human work within the volume of automatically generated content.
Not everyone present shared that alarm. Singer-songwriter Jacopo Ettorre offered a more pragmatic reading, comparing AI to other tools musicians adopted over time, including synthesizers and samplers.
A number for the negotiating table
Economist Will Page, former chief economist at Spotify, offered a figure he uses as a lobbying tool with governments. The total value of global music copyright today stands at $47.2 billion, up from $25 billion when he ran the same calculation a decade ago. According to Page, that is the number to bring into a meeting with a prime minister or a president, not abstract arguments about protecting creativity.
His second point was about negotiating power more than the money itself. Historically, in licensing deals, record labels captured most of the value while music publishers representing composers received what remained. With AI, publishers moved faster than labels to close deals, and as a result they are securing parity arrangements instead of the usual asymmetry. It is a signal that speed to the negotiating table determines who captures value, not just the size of the catalog.
"If our work is creating value for you, then value us"
Visual artist Adelaide Damoah captured the central demand of the day in a single sentence that ended up titling CISAC's own coverage of the event. The claim is simple to state, but it requires something that does not exist in most AI training systems today, which is genuine transparency about what data is used and where it comes from.
That demand connects the GEMA-Suno case, the Darcos Bill, and Franglen's testimony. All three point to the same place. The question is no longer whether AI will coexist with collective management systems. It is under what rules of transparency and remuneration that coexistence will happen. Oron's 1927 analogy serves as a reminder that the argument about technology being so new that old rules cannot apply has been made before, and it did not hold up then either.
Source: CISAC