OpenAI is developing an AI model for music generation, according to the source article, with the goal of producing music from either text or audio prompts. The effort signals a renewed push into music AI after OpenAI's earlier Jukebox experiment in 2020.
What OpenAI Is Building
The model under development is meant to generate music from prompts. Those prompts could be written instructions, or they could come from audio.
One example given in the source is creating a guitar track to accompany a song. That detail matters because it suggests the tool is not only aimed at generating complete musical pieces from scratch. It may also be useful for adding parts to existing ideas, filling in accompaniment, or helping shape a track around something a user already has.
OpenAI is reportedly collaborating with music students at the Juilliard School. Their role, according to the source, is preparing sheet music to help train the model.
That places the work at the intersection of AI generation, musical structure, and human-prepared training material. The source does not describe the model's release timeline, product name, technical design, or training process beyond the reported sheet music preparation.
Why Text and Audio Prompts Matter
A music generation model that accepts both text and audio prompts could support different kinds of creative input. Text prompts let a user describe what they want. Audio prompts let a user begin from sound itself.
The source does not say exactly how those prompts would be handled, but the stated goal is clear: generate music from either form of instruction. That makes the system potentially broader than a tool that responds only to written commands.
The guitar-track example shows the practical direction of the work. A user might not need to describe every musical decision in technical language if an audio prompt can guide the model. At the same time, a text prompt could make the tool accessible to people who know the result they want but do not have a finished musical part ready.
OpenAI is also considering potential uses for this technology in advertising. The source does not provide details about specific advertising products or customers, but the connection is straightforward: advertising often needs music that fits a message, a scene, or a short format.
Competition With Suno and Udio
If OpenAI continues this work, it would move directly into a field already associated with startups like Suno and Udio. The source describes OpenAI's development as putting the company in direct competition with those companies.
That competition would not only be about whether AI can generate music. It would also be about how users prompt systems, how useful the output is, and which kinds of creative workflows the tools can support.
The source does not compare product quality, pricing, availability, or user base. It only establishes that Suno and Udio are current reference points for this market, and that OpenAI's model would be aimed at the same general category of music generation.
For OpenAI, the significance is also strategic. The company has already worked on AI-generated music before, but the source says it has not pursued the technology further until now after introducing Jukebox in 2020. This new effort therefore reads as a return to a field OpenAI had previously explored, not as a first step into music.
The Copyright Question
The music industry is wary of these advances, according to the source. That concern is already visible in legal action involving other AI music companies.
Record labels have filed lawsuits against Suno and Udio, accusing them of possible copyright violations. The source does not provide additional legal details, outcomes, or claims beyond that point.
OpenAI CEO Sam Altman has said that rights holders should eventually share in the revenue. The source notes that he raised this point during the troubled rollout of the Sora app.
However, the source also says it is still unclear how this would actually work. That uncertainty is central to the story. A music generation model can be described in terms of prompts and output, but the business and rights framework around that output remains unresolved in the facts provided.
For rights holders, the open question is not just whether AI music tools can generate useful material. It is also how value, permission, and compensation would be handled if such systems become more widely used.
A Return to an Unsettled Frontier
OpenAI's earlier music project, Jukebox, was introduced in 2020 as an early experiment in AI-generated music. The source says OpenAI had not pursued the technology further until now.
That history gives the current work a different frame. OpenAI is not simply entering a new category because rivals exist. It is revisiting a technical area it had already tested, now at a time when AI music tools are more commercially visible and more legally contested.
The result is a story with two tracks. On one track, OpenAI is developing a model that could generate music from text or audio prompts, with reported help from music students at the Juilliard School and possible advertising uses under consideration. On the other track, the wider industry is still wrestling with copyright concerns, lawsuits involving Suno and Udio, and unanswered questions about revenue sharing for rights holders.
For now, the source leaves key details open. There is no stated launch date, no described user interface, and no explanation of how revenue sharing might work. What is clear is that OpenAI is again working on AI music generation, and that its move would land in a field where creative opportunity and rights questions are already tightly linked.