Andreea Gleeson (CEO, TuneCore) on music distribution in an AI-first world

Creative AI has proven to be a powerful litmus test for several converging trends in the music business — from streaming fraud and the ongoing battle for label market share, to the double-edged sword of democratized music creation and content oversaturation, to a lack of unity in global approaches to copyright and data protections.

At this critical juncture, we hosted Andreea Gleeson, CEO of TuneCore , for a Water Cooler interview to dive into the myriad industry trends swirling around creative AI, through the lens of one of the industry’s most critical functions: Music distribution.

TuneCore is one of the world’s largest open music distribution tools, having paid out over $3 billion to artists to date. The company has also become one of the most vocal music-industry stakeholders when it comes to AI, taking a stance that is equal parts proactive and cautious.

In late April 2023, Denis Ladegaillerie, co-founder/CEO of TuneCore parent Believe, claimed on an earnings call that the organization aims “not to distribute any content that is 100% created by AI,” and would deploy technology to block AI-generated content from being distributed using their tools within the following one to two quarters. Just two months later, in June 2023, TuneCore announced a partnership with Grimes and CreateSafe that would allow any artist to upload songs made with Grimes’ custom voice model to TuneCore for official distribution consideration, with GrimesAI as a featured collaborator.

Most recently, on August 2, 2023 Ladegaillerie presented a more bullish perspective on the role of AI in the music distribution value chain. “We do think that generative AI, and what we’re seeing in the market, will empower every artist to make high quality music,” he said during Believe’s latest earnings call. “We think that for a large part of our business for our automated solutions [division], technology will rapidly be available to elevate the quality of the music, allowing younger artists, developing artists to compete faster in the mid-level market segments and towards the top with higher quality music and accelerate their learning curve to improve their craft.”

Our interview with TuneCore CEO Gleeson dove into the logic and nuance behind these strategic decisions around AI, and the opportunities and challenges at large in embedding generative AI advancements from the bleeding edge into large-scale music-industry infrastructure.

Below is a recap of the main themes from the discussion, which was moderated by Water & Music researcher Yung Spielburg. You can also watch the full interview via the video linked below.

Through the lens of the TuneCore x Grimes partnership playbook, Gleeson outlined three core ethical principles TuneCore is implementing in its approach to AI:

1. Consent — namely, “has the artist said it’s OK to use their IP in the training of a model?” In the case of Grimes, she trained her own custom model, with all training data 100% owned by her.

2. Control — in the context of distribution, “does the original content creator and content owner have the ability to say, ‘Yes, I want this iteration of it, but not this one?’” With the Grimes partnership, TuneCore set up a manual approval process where Grimes’ team could review any submissions that were made outside of the portal that Grimes set up herself. This is admittedly the part of the process where large scalability is still a question mark, as each individual song needs to be reviewed by someone on Grimes’ team. Hence, these processes are only realistically able to protect top artists today.

3. Compensation— “How are you making sure that the original copyright owner is being compensated for the value created in a generative AI song drawing on their IP?” We are early on enough in music AI development that there is no standard playbook for artist compensation yet, but the TuneCore x Grimes partnership has set an initial benchmark of having an artist compensated as the source of the data used to train a widely used voice model, and rewarded based on consumption.

When asked about running this playbook with other artists, Gleeson shared that TuneCore remains cautious and methodical in rolling out their AI strategy more broadly, and wants to get the process right before significantly scaling up. “We’re in talks with CreateSafe about if there are additional artists next, but we want to crawl before we run and really get this process down and feel really good about it,” she said. “And then we can turn up the faucet, so to speak.”

Lowering the barrier to creativity, where more people can express themselves, is a net positive thing at large. However, recent and future advancements in AI that allow for music in certain genres to be generated with little to no human input does create inherent friction between providing value to listeners and providing value to artists. If music that is 100% AI-generated is “good enough” to the point of being commercially successful, companies could arguably generate value for listeners by using AI to provide more choice or customization around a given listening experience, while at the same time competing with traditional artists.

When it comes to the market value of 100% AI-generated music, especially in passive background or ambient contexts, Gleeson’s opinion on the topic is still developing. “I don’t know if I have my full convictions on this,” she said. “There can be a space [for AI] for background music, but it’ll be treated differently by the DSPs.”

That said, Believe has made its public stance clear against 100% automated, AI-generated content when it comes to its place in TuneCore’s business model specifically. Gleeson echoed similar sentiments during our interview when it came to the importance of supporting TuneCore’s artist customer base — framing their investment in AI as part of a wider strategy of using technology to accelerate creativity alongside marketing and distribution:

“Ultimately, what we stand behind as a platform and how we’re leaning into AI is we want to support human-created music, and human use of AI to help enhance human creativity. We want to provide those opportunities for artists to make their music better, and then get it to listeners that have an affinity for their music.”

During our conversation, Gleeson touched upon TuneCore’s mission of providing value to artists to help them grow on the one hand, and providing value to listeners on the other hand.

In our industry-wide survey on music AI from February 2023, oversaturation was the greatest concern amongst producers and artists when it came to how AI would transform the music industry. However, Gleeson posited that more listener choice is not inherently a bad thing if artists are able to find smaller audiences that are more dedicated and committed:

“ It’s not necessarily a bad thing to have more choice for listeners … Think about what’s in your closet today versus what it used to look like. With online shopping, you can have much more micro tastes in what you wear every day. The exact same transformation has happened with music … I don’t think getting more music to more niche listeners is a bad thing, because you can have more super fandom. We’re just not going to have a lot more Beyoncés in the world.”

Later on, Gleeson also posited that new kinds of online fan ecosystems could develop around music AI, the way that Grimes has cultivated her fan and creator community hands-on through her custom platform Elf Tech. “There is very deep fandom happening,” said Gleeson. “I think this is something that can absolutely emerge in an ecosystem that’s not necessarily in the major places people are consuming music today.”