During 2024, the global technology industry invested approximately one billion dollars per day in Generative AI. Models improved at a pace that surprised even the people building them. The ability to generate text, music, images, and video that is indistinguishable from human-made content moved from being a research curiosity to a commercial reality.
Most of the conversation about this has focused on what AI can produce. Far less attention has gone to what it consumes, and what it costs those whose work is being consumed.
The Numbers That Cannot Be Ignored
A CISAC report published in 2024 put concrete figures around a risk that the creative industry had been discussing in general terms. Between 21 and 24 percent of creator incomes in music and audiovisual content could be at risk by 2028, as Generative AI displaces demand for original creative work in certain categories.
The mechanism is not complicated. When a company can generate a music bed for an advertisement using an AI model instead of commissioning a composer, it does. When a streaming platform can generate ambient content at scale for a fraction of the cost of licensed recordings, it will. Each of these use cases individually seems minor. Accumulated across millions of daily decisions, they represent a significant displacement of creator income.
The irony is structural. The AI models producing this content were trained on creative work, most of it without compensation or explicit consent. The output of that training now competes directly with the creators whose work enabled it.
The Vicious Cycle
There is a less obvious risk underneath the income displacement. If the economics of creating original content deteriorate significantly, fewer people will create it. And if fewer people create original content, the AI models that depend on high-quality training data will have less of it to learn from, gradually degrading the quality of their outputs.
This is not a problem that the technology industry can solve by simply generating more AI content to train on. The value of creative work to AI systems comes precisely from the fact that it reflects genuine human experience, emotion, and craft. Synthetic data generated by AI feeding on synthetic data loses that value quickly.
A sustainable future for Generative AI requires a sustainable creative ecosystem. These interests are not opposed. They are structurally interdependent.
The Players Building Solutions
Several ventures are working directly on the attribution and compensation problem, and the capital flowing into this space reflects how seriously it is being taken.
ProRata.ai, founded by Bill Gross with over 25 years of experience launching technology ventures, raised 25 million dollars to build a chat AI platform trained exclusively on content whose creators have given explicit consent and received compensation. Their system includes a reverse engineering algorithm that traces the origin of model outputs, enabling passive income for rights holders based on actual usage.
Vermillio, backed by a 16 million dollar investment led by Sony, developed TraceID, a technology that monitors the web continuously for unauthorized use of protected content and intellectual property, and automates the process of removal requests and royalty collection.
Sureel AI built an attribution algorithm that can analyze AI responses and determine which works influenced them. Stim, the Swedish Collective Management Organization, became one of their first major clients, enabling a licensing model for AI training that is grounded in traceable attribution rather than blanket fees.
Story from PIP Labs is building infrastructure at the protocol level that would make intellectual property natively trackable and compensable within AI systems.
The Strategic Role of CMOs
What is notable about every one of these solutions is that they need exactly what Collective Management Organizations already have: a large base of represented creators, legal standing to negotiate on their behalf, and established infrastructure for collection and distribution.
CMOs are not late to this conversation. They are, in many ways, the natural counterpart to what these platforms are building. An AI company that wants to train on music legally needs to license from someone who represents the rights. The organization best positioned to offer that at scale, and to distribute the resulting revenue to individual creators, is a collecting society.
The organizations that move quickly to understand attribution technology, to develop clear licensing frameworks for AI training, and to position themselves as the institutional bridge between the creative and technology worlds will define what collective management looks like in the next decade.
Those that wait for the frameworks to be finalized without them will inherit whatever was designed in their absence.
The Labs at global.esur works with Collective Management Organizations on the technology and strategy challenges at the frontier of the industry. Get in touch to explore what this means for your organization.