Record labels sue Suno and Udio! Is it good for AI?

It was reported yesterday that the RIAA sued Suno and Udio for copyright infringement based on the songs they used to train their models and the songs they’re producing.

This is good for AI. Hear me out.

I’m not a lawyer, but IP law fascinates me and the legal questions around AI are especially interesting. More importantly, they’re especially pressing, because there’s a huge grey zone in understanding what the legal status of both the models and the output they produce is.

In the beginning it was just some celebrity authors, a few newspapers, and some visual artists filing suit. But now the RIAA has gotten involved and MPA (formerly MPAA) should be entering the fray soon. This should help spur Congress and/or SCOTUS to action to get this resolved.

What needs to be resolved?

The two big questions are whether training AI on copyrighted works constitutes infringement or fair use and how much similarity constitutes infringement. We either need to have language in the law that clarifies these better or the courts will have to do Congress’s job for it… again.

Remember the wording of the constitutional article that gives Congress the power to regulate IP. It’s granted “to promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries.”

Essentially, that pretty much means what Congress says goes and SCOTUS can just clarify ambiguities when it comes to IP. That said, there are some limits on the changes they can make because of our treaties regarding IP with other nations, but that would be for Congress to work out when drafting the law, not for SCOTUS to parse.

In the end, the AI battles’ (visual, prose, audio, code) side benefit will *hopefully* be deeper clarification on infringement via similarity and what constitutes fair use. This would HELP the builders and sellers of AI by giving them a clear and relatively unambiguous path forward. It might not be the path they want, but it would eliminate uncertainties for both them and their customers. A lot of people aren’t using AI because they’re afraid they could either lose the ownership of their AI-produced works or even get sued over the works.

Long term, a clear path forward to create AI that is within the bounds of the law and is legally bulletproof (as much as is possible) benefits the industry. For creators of original work and those using AI-created work, having a clear picture of their rights to those works (ownership, exclusivity, fair use) will benefit artists, AI companies, and consumers of AI services.

How soon can we expect this to be resolved?

It’d just be nice if Congress got off its ass and clarified it in legislation this year instead of waiting for SCOTUS to announce a first half-ambiguous decision in June of 2028 and for it to take until the mid 2030s for it all to be worked out in the courts.

Until it’s clear, the gray zone is creating a gordian knot of toothpaste that will be increasingly difficult to put back in the tube. Congress needs to get off its lazy ass, stop shirking its duties, and do something about it rather than punt it to the courts or wait for a new trade treaty to backdoor in rules the Executive Branch negotiated with the EU or China.