An international music publishing association, the ICMP, has unveiled a significant investigation revealing what it describes as “extensive evidence of serious copyright infringement” by various AI firms. The association claims that these firms have unlawfully utilized copyrighted music from renowned artists such as The Beatles, Elton John, Madonna, and others to train their generative AI models.
This revelation comes in the wake of previous legal actions taken against companies like Udio and Suno by major record labels and independent artists. However, the ICMP’s findings represent the most extensive probe to date into the use of copyrighted music in training generative AI systems within the music industry.
Billboard was given exclusive access to the ICMP’s findings, which were compiled through a thorough examination of public registries, open-source repositories, leaked materials, research papers, and insights from AI experts. The association asserts that it has amassed comprehensive evidence demonstrating the unauthorized use of digital music for AI training and GenAI music, as well as the creation of songwriter and performer likenesses.
The documents shared with Billboard name several prominent AI firms, including Suno, Udio, Microsoft’s CoPilot, Google’s Gemini, OpenAI (developer of ChatGPT), and Midjourney, among others. Of particular note is an alleged admission from OpenAI that its Jukebox music-making app, which debuted in 2020, was trained on music from a vast array of artists, including iconic names like The Beatles, Elton John, and Madonna.
John Phelan, the ICMP’s director general, has characterized the widespread copyright infringement uncovered by the association as “the largest IP theft in human history.” Phelan emphasizes that the unauthorized scraping of musical works from platforms like YouTube, Spotify, and GitHub for AI model training purposes represents a flagrant violation of music publishers’ and songwriters’ rights.
Despite the music sector’s insistence on the necessity of obtaining licenses for legal AI training and GenAI, many tech companies have been caught red-handed engaging in copyright infringement practices. These firms have allegedly been extracting lyric datasets from the internet without authorization, disregarding the rights of the music creators and publishers.
In response to previous legal challenges, some AI firms have cited the “fair use” doctrine to justify their actions, claiming that their use of copyrighted material falls within the bounds of permissible usage. However, the ICMP’s investigation challenges this defense, asserting that the scale of copyright infringement observed is unprecedented and cannot be excused under fair use provisions.
Billboard reached out to all the implicated tech companies for comment, but they either declined to respond or did not provide a statement. The music industry continues to grapple with the implications and consequences of AI firms using copyrighted music without proper authorization, raising critical questions about intellectual property rights and ethical AI practices.
📰 Related Articles
- US Law Firms Urged to Prioritize AI Training for Competitiveness
- Financial Firms Navigate AI Regulations for Innovation and Compliance
- Builder.ai’s Collapse Exposes Risks of AI Investment Frenzy
- e.l.f. Beauty & Pinterest Launch AI Color Analysis Tool
- Xerox Versalink C7025 Printer Vulnerability Exposes Network Security Risks