AI music company Suno has a $120 billion target, however is copyrighted music involved?

AI music company Suno has emerged as a large player in AI music generation which charges users a subscription fee to create full songs from simple inputs like text prompts or lyrics. While Suno has become well-known for advanced listening compositions, there are several concerns regarding its training practices. Despite leveraging OpenAI’s API for lyrics generation, Suno’s failure to disclose its training data sources raises suspicions of potential copyright infringement.

Music AI companies have become significantly popular in this day and age attracting millions of users and commanding valuations with their offerings of AI models capable of generating text, images, code, speech, and more. In a recent article via Music Business Worldwide, Newton-Rex takes a deep dive into AI music company Suno’s 120 billion dollar evaluation and if copyright music may in fact be involved.

As many of these models are trained on vast amounts of copyrighted content without proper permissions or compensation, resulting in significant backlash from creators and media entities alike. Among these enterprises, Suno emerges as a significant company for AI music generation, charging users a subscription fee to create full songs from simple inputs like text prompts or lyrics. While Suno’s has now become well known for advanced listening habits, there are several concerns related to its training practices. Despite leveraging OpenAI’s API for lyrics generation, Suno’s failure to disclose its training data sources raises suspicions of potential copyright infringement. Statements from Suno’s investors further suggest legal risks, hinting at the possibility of training on copyrighted material without proper authorization. Newtown shares:

“I, and others, have found that Suno regularly outputs music that closely resembles copyrighted material. This is true across musical style, melodies, chord sequences, instrumental parts and lyrics. In this post, I will share some examples, and evaluate what they mean. I want to stress that I don’t think this kind of output resemblance is required for copyright to have been infringed: I think that, if the model was trained on copyrighted work without a license, copyright has been infringed, whether the model generates exact copies of the training data or not. The examples I include in this post are rather intended to help point towards what the model may have been trained on.”