Suno is switching to watermarking technology to identify AI-generated music. Here are the platform’s new transparency and security steps.
Suno, an AI-based music production platform, announced the development of a new watermarking system to make music created on its platform identifiable. The company stated that it took this step in light of intense pressure from the music industry and ongoing legal processes.
Suno co-founder and CEO Mikey Shulman announced in a blog post that new transparency tools will be implemented. This system includes a tagging system that can show whether a piece of music was produced on Suno on other platforms as well.
Watermarks and Download Limits Coming
Shulman stated that the company will use new audio watermarking and fingerprinting technologies to become more resistant to AI-related fraud attempts. This approach works with a logic similar to Google’s SynthID system, allowing the source to be traced even if the content is edited or compressed.
In addition to the watermarking system, the platform will also introduce new download limits to prevent the mass distribution of music. This decision comes after major record labels such as Sony Music, Universal Music Group, and Warner Music Group demanded the removal of AI-generated music that has invaded the charts.
Copyrights and Legal Processes
Companies such as Suno and Udio have faced numerous copyright lawsuits, defendants of training their models with massive amounts of music data without permission. Last November, Warner Music reached a licensing agreement with Suno, ending the existing legal dispute.
However, legal pressure on the platform continues, and recently a German court upheld Gema’s copyright infringement claims. The Munich-based court ruled that Suno trained its systems on protected music to which it did not own the rights; the company stated that it could appeal the decision.
Suno announced that it is collaborating with data companies such as Audible Magic and Musixmatch to prevent misuse of the platform. Thanks to these collaborations, uploaded audio files and lyrics will be scanned for potential infringements.
Do you think marking AI-generated music in this way is a sufficient step to solve copyright problems?