Sony Group has introduced a pioneering technology to identify the copyrighted music used in AI-generated songs. This innovation aims to address a major challenge in the entertainment industry: ensuring that songwriters and artists receive proper compensation when their work is used without permission by AI models.
The system evaluates AI-created tracks to determine the contribution of original artists, for example, identifying if a song is “30% derived from The Beatles and 10% from Queen.”
When AI developers cooperate, Sony can directly access their models to trace training data. If there’s no cooperation, the tool estimates origins by comparing the output with music databases. Amidst a surge of AI-generated content, including viral songs with “deepfake” vocals of stars like Drake and The Weeknd, Sony’s new tech aims to promote fair revenue sharing.
With extensive music catalogs, including rights to half of Michael Jackson’s estate, Sony sees this as essential for fair royalty collection from AI companies, similar to current methods for radio and streaming.
Developed by Sony AI, this technology also extends to visual styles, helping prevent AI from mimicking specific artistic styles, such as Japanese anime or Studio Ghibli films.
While Sony envisions a future in which content companies utilize this tool to negotiate licensing deals, obstacles still exist. The company has acknowledged that it has not yet set a launch date.
Additionally, industry experts caution that AI developers, who frequently prioritize model performance over intellectual property rights, may be hesitant to adopt technology that could expose them to copyright claims.