Is OpenAI building an AI music generator?
Reports hint at a ChatGPT-powered tool that could create music from scratch, but the industry may not be on board just yet.
OpenAI’s music interest
It looks like OpenAI is ready to dive into the world of music creation. According to The Information, the company behind ChatGPT is reportedly developing its own AI music generator that could allow users to create music directly from text or audio prompts.
Sources close to the project say that OpenAI has teamed up with students from New York’s Juilliard School to bring the vision to life. These students are said to be annotating music scores to help train the AI into a model capable of generating realistic-sounding music.
What could the tool actually do?
Right now, there’s limited details about the specifics of the tool or when it could be launched. Reports suggest OpenAI imagines a tool that can generate instrumental accompaniments, or add music to videos. Essentially, a creative assistant for musicians, producers, and content creators alike.
As always, there’ll likely be a legal catch. As Music Ally points out, training AI models on annotated music scores could raise legal eyebrows. While OpenAI could rely on out-of-copyright works, a truly capable model would likely need exposure to a much broader range of music. That’s where publishers and rights holders may step in with lawsuits.
OpenAI and music: Not for the first time
This isn’t OpenAI’s first rodeo into AI music generation. Back in 2019, the company launched MuseNet, an AI that could combine styles from different artists. This was followed by Jukebox in 2020, which could generate songs in the style of legends like Elvis Presley or Frank Sinatra.
Of course, these experiments also raised red flags about copyright and artist likeness. OpenAI’s CEO Sam Altman acknowledged these issues in 2023, saying that creators should have control over how their work and likeness are used.
“Creators deserve control over how their creations are used, and what happens beyond the point of them releasing it into the world. We think that content creators, content owners, likenesses: people totally deserve control over how that’s used, and to benefit from it.”
Sam Altman, OpenAI’s CEO (via Music Ally)
Yet, as we recently saw with OpenAI’s video model Sora, concerns remain about AI’s potential to replicate people and creative works without consent.
The industry’s evolving relationship with AI
Despite the legal grey areas, the music industry seems to steadily be embracing AI, or at least learning to work with it. Just last week, we reported on the recent $2 billion valuation of Suno (another AI music generator), showing that investors see huge potential in this space, even if lawsuits are mounting.
Even OpenAI is already working with Spotify across two different features: Spotify’s AI DJ tool, and a recent full integration of Spotify within ChatGPT. Speaking of Spotify, the streaming giant is now working with major labels and independent partners to build “responsible” AI music tools– with a clear emphasis on proper licensing and compensation for artists.
What this means for musicians
If OpenAI does launch a full AI music generator, it could shake things up both creatively and legally. While simple tools that can act as a creative assistant are more welcomed, full-scale music generators that can compose and perform entire tracks are much more controversial.
For artists, the focus is still on having control and compensation for their works, for both the AI training and its outputs. For the wider industry, this move could signal another step towards AI-integrated creativity, or further inspire the ongoing debate about ownership and originality in AI music.