What the RIAA lawsuits against Udio and Suno mean for AI and copyright

Udio and Suno are not, despite their names, the hottest new restaurants on the Lower East Side. They are AI startups that allow people to generate stunningly real-sounding music – complete with instrumentation and vocal performances – from prompts. And on Monday, a group of major record labels sued, alleging copyright infringement “on an almost unimaginable scale,” claiming that the companies can only do this because they have illegally ingested huge amounts of copyrighted music to train your AI models.

These two lawsuits contribute to a growing pile of legal headaches for the AI ​​industry. Some of the most successful companies in the industry have trained their models with data acquired through unauthorized scraping huge amounts of information coming from the Internet. ChatGPT, for example, was initially trained in millions of documents collected from links posted on Reddit.

These processes, led by the Recording Industry Association of America (RIAA), address the music rather than the written word. But how The New York Times‘ against OpenAI, they raise a question that could reshape the technology landscape as we know it: can AI companies simply take whatever they want, turn it into a product worth billions, and claim it was fair use?

“That’s the key issue that needs to be resolved, because it cuts across all sorts of different industries,” he said. Paulo Faklerpartner at the law firm Mayer Brown specializing in intellectual property cases.

Both Udio and Suno are relatively new, but they have already been a huge success. Suno was launched in December by a Cambridge-based team that previously worked for Kensho, another AI company. It quickly formed a partnership with Microsoft that integrated Suno with Copilot, Microsoft’s AI chatbot.

Udio was released just this year, raising millions of dollars from heavyweights in the world of technology investments (Andreessen Horowitz) and the music world (Will.i.am and Common, for example). The Udio platform was used by comedian King Willonius to generate “BBL Drizzy,” the Drake diss track that went viral after producer Metro Boomin remixed it and released it to the public so anyone could rap.

The RIAA’s lawsuits use cavalier language, saying this litigation aims to “ensure that copyright continues to encourage human invention and imagination, as it has for centuries.” That sounds good, but ultimately the incentive we’re talking about is money.

The RIAA says generative AI poses a risk to record labels’ business model. “Instead of licensing copyrighted sound recordings, potential licensees interested in licensing such recordings for their own purposes could generate AI-like sound at virtually no cost,” the lawsuits state, adding that such services could “[flood] the market with ‘copycats’ and ‘lookalikes’, thereby destroying an established sample licensing business.”

The RIAA is also seeking damages of $150,000 for infringing work, which, given the enormous pool of data that is typically used to train AI systems, is a potentially astronomical number.

The RIAA’s filings included examples of music generated with Suno and Udio and comparisons of their musical notation with existing copyrighted works. In some cases, the generated songs had similar short phrases — for example, one began with the line sung “Jason Derulo” in the exact cadence that the real-life Jason Derulo begins many of his songs. Others had extended sequences of similar notation, as in the case of a track inspired by Green Day’s “American Idiot.”

That looks Pretty damning, but the RIAA isn’t claiming that these specific soundtracks infringe copyright – instead, it’s claiming that the AI ​​companies used copyrighted music as part of their training data.

Neither Suno nor Udio have made their training datasets public. And both companies are vague about the sources of their training data — although that’s par for the course in the AI ​​industry. (OpenAI, for example, has dodged questions about whether YouTube videos were used to train their Sora video model.)

The RIAA filings note that Udio CEO David Ding said the company trains with “best quality” music that is “publicly available” and that a Suno co-founder wrote on the official Suno Discord that the company trains with a “mix of proprietary music and public data.”

Fakler said including examples and notation comparisons in the lawsuit was “crazy,” saying it went “way beyond” what would be necessary to claim legitimate grounds for a lawsuit. On the one hand, record companies cannot own the composition rights to the songs allegedly ingested by Udio and Suno for training. Instead, they own the copyright to the sound recording, so showing similarity in musical notation doesn’t necessarily help in a copyright dispute. “I think it was really designed for optics for public relations purposes,” Fakler said.

Additionally, Fakler noted, it’s legal to create a similar audio recording if you have the rights to the underlying music.

When reached for comment, a Suno spokesperson shared a statement from CEO Mikey Shulman stating that its technology is “transformative” and that the company does not allow notices that name existing artists. Udio did not respond to a request for comment.

But even if Udio and Suno used the record companies’ copyrighted works to train their models, there’s one big question that could override everything else: Is this fair use?

Fair use is a legal defense that allows the use of copyrighted material in creating a significantly new or transformative work. The RIAA argues that startups cannot claim fair use, saying that the results from Udio and Suno are intended to replace real recordings, that they are generated for commercial purposes, that the copying was extensive and not selective, and finally that the resulting product represents a direct threat to the record companies’ business.

In Fakler’s view, startups have a solid fair use argument, as long as the copyrighted works were only temporarily copied and their defining characteristics were extracted and abstracted into the weights of an AI model.

“That’s how computers work – they have to make these copies, and the computer analyzes all this data so it can extract the non-copyrighted material,” he said. “How do we construct songs that will be understood as music by the listener and that have several characteristics that we commonly find in popular music? It’s about extracting all of that, just like a musician would learn these things by playing music.”

“In my opinion, this is a very strong fair use argument,” Fakler said.

Of course, a judge or jury might not agree. And what is discovered in the discovery process – if those processes get there – could have a big effect on the case. Which music tracks were recorded and how they ended up in the training set can be important, and details about the training process can undermine a fair use defense.

We are all in for a long journey as the RIAA lawsuits, and others like them, make their way through the courts. From text and photos to sound recordings, the question of fair use looms over all of these cases and the AI ​​industry as a whole.