Suno admits using music from YouTube to train AI models

Suno’s latest confession came not from a press conference but from a court filing in Boston. On September 1, the company admitted that its AI models had been trained on music pulled from YouTube, a fact that landed it squarely in the crosshairs of the industry’s biggest labels. The lawsuit, filed by Universal Music Group, Sony Music, and Capitol Records, argues that Suno used software such as YT‑DL and YT‑DLP to bypass YouTube’s technical safeguards and download copyrighted recordings without permission.

The story begins in a quiet office in Boston, where a team of engineers was fine‑tuning algorithms that could generate original songs from scratch. In the background, a playlist of YouTube tracks spun on a loop, providing the raw material that the models would learn from. The engineers, unaware of the legal quagmire that would later unfold, fed the AI with thousands of hours of music, hoping to teach it the nuances of melody, rhythm, and lyricism. This practice, now under scrutiny, highlights a broader industry dilemma: where does the training data for AI music come from, and who owns the rights to it?

The lawsuit alleges that Suno’s use of YT‑DL and YT‑DLP constituted a deliberate attempt to circumvent YouTube’s protections. The labels claim that this activity amounts to a breach of copyright law, as the AI models were trained on copyrighted material without the necessary licenses. In response, Suno’s legal team argues that the labels themselves are engaging in anticompetitive behavior by monopolising the production and commercialisation of music. They further claim that the AI’s output is a new, non‑infringing product, thereby falling under the doctrine of fair use.

Last year, a similar battle unfolded when Warner, Sony Music Entertainment, and Universal Music Group sued Suno and another AI company, Udio, over alleged copyright violations. The lawsuit was eventually settled when Warner signed a licensing deal with Suno. That agreement introduced a licensed model into Suno’s offerings, allowing users to purchase AI‑generated songs that were built on licensed content. The deal also ensured that artists who opt into AI collaborations would be compensated fairly and retain full control over their music, likeness, and other copyright details.

When Jason Isbell filed a lawsuit against Suno this month, he accused the software of enabling users to replicate an artist’s identity and musical style without consent. This case underscores a deeper concern: the potential erosion of artistic ownership algorithms can mimic styles with alarming precision. For musicians, the fear is not just about lost revenue but about the dilution of their creative voice.

The dispute is not isolated. Last month, Suno lost a copyright infringement case to the German licensing agency GEMA, which argued that the company had used songs such as Boney M’s “Daddy Cool” and Lou Bega’s “Mambo Jambo” without permission. These incidents collectively paint a picture of an industry grappling with the intersection of technology and intellectual property.

As a writer who has watched the music scene evolve from vinyl to streaming, I find this clash both unsettling and inevitable. The promise of AI music is undeniablenew sounds, faster production, democratised creation. Yet the legal framework has not yet caught up with the speed of innovation. If Suno’s admission is a warning, it is also a call for clearer guidelines that protect artists while fostering technological progress.

Looking ahead, the outcome of this lawsuit could set a precedent for how AI companies source training data. Should the court side with the labels, it may force firms like Suno to overhaul their data acquisition practices, potentially slowing the pace of AI‑generated music. If the court defers to Suno’s fair‑use defense, the industry may see a surge in unlicensed training, raising ethical questions about ownership and compensation. Either way, the conversation around AI and music is poised to intensify, demanding a balance between innovation and respect for artistic rights.