OpenAI is Developing an AI Music Generator with Help from Juilliard

According to The Information, OpenAI is widening its generative repertoire beyond text, image and video to model-driven music creation. The article says the company is developing a tool capable of generating music from text or audio prompts, with Juilliard students reportedly helping prepare training data by annotating musical scores. These annotated datasets are intended to train AI models to understand musical structures and elements, according to sources familiar with the matter.

The piece suggests potential use cases such as automatically composing guitar accompaniments for existing vocal tracks or adding fitting background music to videos. It notes that several startups, including Suno and ElevenLabs, have already rolled out AI music-generation services, though the quality of outputs remains inconsistent and raises debates about copyright and creative potential. The report was published on 27 October 2025.