Charlie Puth is going all in on AI.
Moises, a music-tech company, announced Wednesday that Puth, the Grammy-nominated singer, has been tapped as its chief music officer, a role that entails guiding its creative and product direction.
“Every musician I know is using Moises, and I’ve been using it in my own creative process for years,” Puth said in a statement. “It opens up possibilities that used to take hours or expensive studio setups, whether that’s isolating vocals to study technique or experimenting with arrangements in real time.”
For several years, there have been contentious debates in Hollywood over regulating AI. Many creatives remain wary about AI technology, especially when content generation capabilities are involved.
But in recent months, more AI companies have pushed to strike licensing deals with talent across sectors — from film and TV to video games — to avoid blowback from artists who worry that their voices and likenesses are being used without their consent.
Notably, the AI voice generation platform ElevenLabs teamed up with the likes of actors Matthew McConaughey and Michael Caine to license their famous voices for generative audio uses. Singer Liza Minnelli also collaborated on an AI-generated album made by the company.
Puth has been a public supporter of experimenting with AI.
In 2023, he was among a handful of celebrities who became the first to sign onto an AI music partnership. Puth collaborated with Google to make his voice available for AI-generated YouTube Shorts soundtracks.
He has also been a longtime user of Moises, having recently partnered with it to launch a “Jam Sessions” contest for his fans.
“AI, when done right, isn’t here to replace musicians,” Puth said in his statement about his new role.
Moises, founded in 2019 as an AI-powered service marketed to musicians, is primarily used to isolate vocals and instruments, which is often necessary in remixing or sampling songs. The company — which has 70 million users worldwide — also touts tools to detect and change song keys or to detect and generate the correct guitar chords for songs.
Last year, it made further forays into generative AI by building an AI-augmented music studio that allows users to generate song stems, the individual music files that make up tracks, by inputting text prompts or their own audio snippets as references.
Co-founder and CEO Geraldo Ramos said he wants to differentiate Moises from AI music generators “where you can just bang the keyboard and get a full song.”
“You can generate building blocks of music with text. So, for example, you can start with a guitar and say, ‘I want a bass here that’s funky and groovy.’ And then it creates a bass that fits your initial input,” Ramos said. “But you can’t just go from zero to a full song with only one prompt. We don’t offer that solution.”
Ramos said Puth also used the platform ahead of his Super Bowl national anthem performance.
Puth loaded tracks into Moises to practice and also used it to experiment with different keys and other elements for his upcoming album, “Whatever’s Clever!”
In recent months, record companies have also tried to embrace the rise of AI in music by negotiating deals that compensate human artists.
Late last year, Universal Music Group and Warner Music Group signed major licensing deals with the AI music studios Stability AI and Udio — with both music labels settling their copyright infringement lawsuits against Udio. Around the same time, UMG, WMG and Sony Music Entertainment also announced separate AI licensing agreements with Klay, a small music technology company based in Los Angeles.
Many such licensing deals have been born out of artists’ continued pushback against and concern about the use of their work and likenesses to train AI models without permission or compensation. In 2024, more than 11,000 creative professionals signed an open letter calling for prohibiting using human art to train artificial intelligence without permission.
Moises has also struck licensing deals with singers for its own AI voice models. Ramos said the company has them record their voices in a studio for around a week, paying for their time and licensing the vocals they produce. Then, he said, part of the platform’s subscription revenue is divided among those singers on a recurring basis.
As AI-generated music proliferates online, some listeners have lamented its quiet incorporation into streaming services. Last year, around 6,300 Spotify users voted in a live poll for the platform to “introduce a clear label for AI-generated songs and provide an option to filter them out entirely.”
The industry has also struggled with the technology. Harvey Mason Jr., president and CEO of the Recording Academy, told Billboard in December that how the Grammys handles AI-generated or AI-assisted music is “the toughest part of my job.”
He said he has seen uses ranging from someone generating an entire track to someone producing an entire song on their own and then supplementing small tidbits with AI.
“These people that are professionals are generally somewhere in the middle, where they’re using it as a tool,” Mason said. “They’re unlocking something when they’re stuck on a lyric, or they’re trying to find 15 things that rhyme with ‘this.’ Also, people are using it as inspiration, not just taking what it gives you. They are just using it as a launch point.”
He noted that using AI doesn’t necessarily make an entry ineligible for the Grammys, but it requires hopeful nominees to “choose the right categories to be considered in.”
Ramos said he believes much of the pushback comes from people’s annoyance that AI models’ outputs are diluting a market once filled with the creations of human artists. He said that because Moises doesn’t generate full songs, it has largely avoided that kind of backlash.
“I think we have a lot of less resistance on that front because of the nature of the products that we’re doing. So that’s why we were able to partner with artists like Charlie,” Ramos said. “I think we’re well-positioned in a different space compared to these generators.”
Puth has also said he doesn’t believe AI is “ever going to completely take over human-made music, mainly because what makes human-made music so special are its imperfections.”
He said in a social media video last year that AI tends to erase human mistakes that would otherwise “add to the vibe,” saying too much technical perfection can take away from the music.
“AI is never going to wipe us off the planet creatively. Like every new piece of technology that comes around every decade, we humans have to learn how to work with it to make music that no one’s ever heard before,” he said. “We are all imperfect beings, but that’s what makes art relatable.”
