Musicians Raise Copyright Concerns Over AI Training

Musicians and the American Federation of Musicians are raising concerns regarding the use of copyrighted songs to train artificial-intelligence models [1, 2].

This movement highlights a growing conflict between generative technology and intellectual property rights. As AI models become more capable of mimicking human artistry, the music industry faces a fundamental shift in how creative work is valued and protected.

Artists said the practice raises critical questions about the role of AI in the music business [1, 3]. The core of the dispute centers on whether using existing recordings to teach a machine constitutes copyright infringement [1, 3].

Beyond the legal definitions of ownership, musicians are concerned about unfair compensation [1, 3]. They said that AI companies are benefiting from decades of human creativity without providing payment or seeking permission from the original creators.

Industry representatives said that the current trajectory of AI development could undermine the livelihood of professional musicians. By automating the creation of melodies and arrangements based on copyrighted data, the technology may replace the very artists it relies upon for training ^[1].

Legal challenges are emerging as a primary tool for these creators. The American Federation of Musicians has been among the groups voicing opposition to the unauthorized use of music in these datasets [1, 2].

These disputes are part of a broader global conversation regarding the ethics of large-scale data scraping. The music industry is now seeking a framework where AI development does not come at the expense of artist rights ^[3].