Composing the Future: AI's Role in Transforming Music

Executive Summary

AI music is catching on, and improvements are rapid. Suno and Udio have raised hundreds of millions of dollars and already have millions of users. Even just comparing four years ago to now, differences in quality are stark, AI music generation arguably seeing more rapid progress than any other area.

The music space is likely to become more competitive. Many artists are already struggling to break out, and AI music will likely further increase competition. AI music is expanding beyond streaming into live performances, with AI-infused acts already proving profitable and transformative for the industry.

Our relationship to music is likely to change. We might soon find ourselves in a world where recommendation systems know what we’ll next like better than ourselves, a world where music listening becomes more active and where we’re interacting with and changing songs as we listen. 

Three policy debates loom large:

1. How should we react to the use of copyrighted works to train the models? AI music companies claim their use should be exempt under fair use standards, arguing the outputs are “transformative”. To decide it’s not fair use would risk pushing innovation overseas and creating monopolies out of the labels, but would likely help current artists by slowing things down. Compromises might exist, like increasing data transparency. 
2. Should there be a federal right of publicity (ROP)? One solution to those using AI to copy the styles of others is to create a federal ROP, which could give artists a property right in their voice and reproductions of it. Most states have a ROP, and instituting it at the federal level could create needed clarity, but would require deciding thorny issues like how to balance a ROP with the First Amendment.
3. Should AI music be copyrightable? The US Copyright Office has thus far responded “no”, but their proposed solution, to evaluate submissions on a case by case basis, isn’t sustainable. The most sensible solution is likely to allow for the copyright of AI music.

‍We can’t just focus on copyright. Further solutions will be needed to navigate the changes from AI music. One promising solution is to switch to a user-centric model of payouts for streaming revenue, which gives listeners greater control over where their money goes and benefits smaller, emerging artists.

Read the full report here.