Suno Says It Will Add Watermarks to AI Music After Lawsuits, German Ruling, and a Data-Scraping Hack

Suno, one of the largest AI music generation platforms, says it will start watermarking everything it produces. The company's CEO and co-founder, Mikey Shulman, announced the plan in a company blog post, framing it as necessary to meet what he called "emerging industry standards" for labeling AI-generated content.

Under the plan, every track Suno's models generate will carry an inaudible, embedded code in the waveform. Platforms like Spotify or YouTube could use a detector trained on that signature to identify AI tracks and choose to label them or block them outright. Shulman says the watermarking technology is "durable and resistant to tampering" and won't degrade audio quality, according to Ars Technica's reporting on the announcement.

Shulman did not specify whether Suno will build its own detection system or license an existing one. Google has been pushing its SynthID watermarking tool as an industry option and recently opened it up for other companies to license. Google says SynthID has already been used to label 60,000 years' worth of audio generated by its Gemini models and more than 100 billion images and videos, according to Ars Technica.

Why now

Suno isn't doing this in a vacuum. The company is being sued in the United States by both Universal Music Group and Sony Music for copyright infringement. A German court separately found that Suno violated the country's music licensing laws, according to Ars Technica's report. Both legal fights could result in significant financial penalties.

On top of that, Suno is dealing with fallout from a hack in late 2025 that exposed how the company trained its AI models. According to Ars Technica, the breach revealed that Suno had scraped content from YouTube, Deezer, and other platforms without disclosure, and the exposed data may have included private information belonging to millions of users. Suno did not disclose the hack when it happened. The company is now facing a lawsuit in Massachusetts over that incident.

Taken together, that's a rough run for a company trying to position itself as a legitimate tool for musicians rather than, in Shulman's own words from the blog, a way to generate "unlimited slop" for streaming platforms. Shulman argues AI "can't replace the human experiences, emotions, and imperfections that make music meaningful," and says Suno already has internal rules meant to guide responsible use of its tools.

What watermarking actually solves, and what it doesn't

There's a legitimate case for watermarking. Streaming services are being flooded with AI-generated music, much of it low-effort and designed to farm royalties or game algorithmic playlists. Giving platforms a reliable way to detect and label that content is a real improvement over the current free-for-all, where listeners often can't tell if a track was made by a human or a machine.

But the fix has real limits. If someone manages to crack Suno's watermark, every track generated up to that point could lose its label retroactively, according to Ars Technica. Watermarking schemes across the AI industry, in images and video as well as audio, have faced repeated attempts to strip or spoof them.

There's also the coverage gap. Suno adopting watermarks does nothing about competing platforms that have no such plans, or about open-source AI music models that let anyone generate unlabeled tracks on their own hardware. A determined bad actor doesn't need Suno's tools if unwatermarked alternatives exist elsewhere.

Watermarking doesn't touch the actual legal fight. The lawsuits from Universal and Sony, and the German licensing ruling, are about whether Suno had the right to train its models on copyrighted material in the first place. Labeling the output as AI-generated doesn't resolve the underlying question of whether that output was built on stolen work. Neither does it address the unresolved Massachusetts case over the undisclosed hack and what user data may have been exposed.

Ars Technica's framing treats the watermark announcement largely as a public-relations pivot, noting Suno is "looking for ways to change the conversation" after a string of bad headlines. That reading aligns with the timeline: the watermarking plan surfaces after the lawsuits, the German ruling, and the hack disclosure, not before.

The open questions now are practical ones. Will Suno partner with Google's SynthID or build something proprietary? Will major platforms actually adopt Suno's detector and act on it, given they have no obligation to? And does labeling AI tracks change anything for the copyright lawsuits still working through U.S. and German courts, where the real financial exposure sits? Suno hasn't said when the watermarking rollout begins.