Manifesto:
Attribution in Generative AI Music Models
Generative AI and music innovations are happening so quickly these days. As this is all unfolding, we think that it’s a particularly important time to create bridges between technology companies and rights holders. These two worlds, though interconnected, often work at completely different speeds. They sometimes have different values, and prioritize differently.
At Musical AI, we believe in two Big Things. We believe that it's important (to all of humanity) to unlock the full potential of AI. We also think that it is critical to preserve the integrity of artists and their works. While there is a false perception that these are competing assertions that could lead to a lose-lose middle-ground scenario—possibly driven by greed and a scarcity mindset—we see a potential win-win by both working together from early on. A middle ground must be found where innovation and artistic integrity coexist, benefiting everyone involved.
Digital colonialism is the risk we face. The unchecked appropriation of human creative output by technology companies without consent, credit, or compensation; actions that mirror the exploitative practices of historical colonialism.
This practice undermines the rights of artists, authors and other creators, and also erodes the cultural heritage embedded within those works. Intellectual property must be respected to ensure that those who contribute their artistry to AI models are recognized and rewarded appropriately.
Conversely, technology companies need easy access to high-quality human creative output to train and refine their groundbreaking AI models. We can't make great AI without great art feeding it.
Innovation works when the environments have few obstacles, and where barriers to entry are low. Without easy access to rich and diverse datasets, the advancement of AI technologies that approach their products in legal ways are at risk from being surpassed by the alternatives.
It's a two-way street.
Both industries must engage in open dialogue and collaborative efforts to develop frameworks that balance the needs of artists and innovators.
And so, this is what we believe in:
We must reject digital exploitation. We must recognize that unauthorized use of creative content constitutes a modern form of colonialism.
We must establish equitable licensing agreements. We must create standardized, fair, and transparent licensing models that allow AI companies to access music content legally while ensuring rights holders are justly compensated.
We must implement robust attribution systems. We must facilitate the use of technologies and practices that accurately attribute AI-generated outputs to the original creators whose works influenced them, preserving the lineage of artistic contribution.
We must promote the fairly trained use of data. We must ensure that all music used in AI training is sourced in a fair way, with the explicit consent of rights holders.
We must facilitate mutual growth. We must encourage an environment where technological leaps and artistic integrity coexist.
We must give space for innovation without sacrificing respect for intellectual property.
By finding this middle ground, we can foster an ecosystem where creativity and technology enhance one another without replicating the exploitative dynamics of the past. Rights holders benefit from new revenue streams and wider recognition, and AI companies innovate within a legal framework that respects the origins of the data that powers their businesses. It’s a win-win.
We believe in a future where both artists and technologists thrive. And so we’ve decided to help build it.
