The RIAA’s lawsuit against generative music startups will be the bloodbath AI needs

Like many AI companies, Udio and Suno relied on large-scale theft to create their generative AI models. This they have as much as admitted, even before the music industry’s new lawsuits against them have gone before a judge. If it goes before a jury, the trial could be both a damaging exposé and a highly useful precedent for similarly unethical AI companies facing certain legal peril.

The lawsuits were filed Monday with great fanfare by the Recording Industry Association of America, putting us all in the uncomfortable position of rooting for the RIAA, which for decades has been the bogeyman of digital media. I myself have received nastygrams from them! The case is simply that clear.

The gist of the two lawsuits, which are extremely similar in content, is that Suno and Udio (strictly speaking, Uncharted Labs doing business as Udio) indiscriminately pillaged more or less the entire history of recorded music to form datasets, which they then used to train a music-generating AI.

And here let us quickly note that these AIs don’t “generate” so much as match the user’s prompt to patterns from their training data and then attempt to complete that pattern. In a way, all these models do is perform covers or mashups of the songs they ingested.

That Suno and Udio did ingest said data is, for all intents and purposes (including legal ones), unquestionably the case. The companies’ leadership and investors have been unwisely loose-lipped about the copyright challenges of the space.

They have admitted that the only way to create a good music generation model is to ingest a large amount of high-quality music, much of which will be copyrighted. It is very simply a necessary step for creating machine learning models of this type.

Then they admitted that they did so without the copyright owners’ permission. Investor Brian Hiatt told Rolling Stone just a few months ago:

Honestly, if we had deals with labels when this company got started, I probably wouldn’t have invested in it. I think that they needed to make this product without the constraints.

Tell me you stole a century of music without telling me you stole a century of music, got it. To be clear, by “constraints,” he is referring to copyright law.

Last, the companies told the RIAA’s lawyers that they believe swiping all this media falls under fair use doctrine — which fundamentally only comes into play in unauthorized use of a work. Now, fair use is admittedly a complex and hazy concept in idea and execution. But a company with $100 million in its pockets stealing every song ever made so it can replicate them in great deal and sell the results: I’m not a lawyer, but that does appear to stray somewhat outside the intended safe harbor of, say, a seventh-grader using a Pearl Jam song in the background of their video on global warming.

To be blunt, it looks like these companies’ goose is cooked. They clearly hoped that they could take a page from OpenAI’s playbook, secretly using copyrighted works, then using evasive language and misdirection to stall their less deep-pocketed critics, like authors and journalists. If by the time the AI companies’ skulduggery is revealed, they’re the only option for distribution, it no longer matters.

In other words: Deny, deflect, delay. Ideally you can spin it out until the tables turn and you make deals with your critics — for LLMs, it’s news outlets and the like, and in this case it would be record labels, which the music generators clearly hoped to eventually come to from a position of power. “Sure, we stole your stuff, but now it’s a big business; wouldn’t you rather play with us than against us?” It’s a common strategy in Silicon Valley and a winning one, since it mainly just costs money.

But it’s harder to pull off when there’s a smoking gun in your hand. And unfortunately for Udio and Suno, the RIAA included a few thousand smoking guns in the lawsuit: songs it owns that are clearly being regurgitated by the music models. Jackson 5 or Maroon 5, the “generated” songs are just lightly garbled versions of the originals — something that would be impossible if the original were not included in the training data.

The nature of LLMs — specifically, their tendency to hallucinate and lose the plot the more they write — precludes regurgitation of, for example, entire books. This has likely mooted a lawsuit by authors against OpenAI, since the latter can plausibly claim the snippets its model does quote were grabbed from reviews, first pages available online and so on. (The latest goalpost move is that they did use copyright works early on but have since stopped, which is funny because it’s like saying you only juiced the orange once but have since stopped.)

What you can’t do is plausibly claim that your music generator only heard a few bars of “Great Balls of Fire” and somehow managed to spit out the rest word for word and chord for chord. Any judge or jury would laugh in your face, and with luck a court artist will have their chance at illustrating that.