By: Sarah McLaughlin
With more than 551 million users, Spotify, one of the world’s most popular audio streaming services, is making a significant entrance into the AI arena by trying to reshape the user experience. In February of 2023, Spotify released their new AI DJ—a “personalized AI guide that knows your music taste so well that it can choose what to play for you.” Using OpenAI technology, the DJ draws on the expertise of Spotify’s music editors combined with a user’s listening history to build a platform that offers a unique listening experience tailored exactly to the users music preferences. This feature seemed to announce Spotify’s approval in embracing AI within the platform.
Shortly after, on April 4th, a song titled “Heart on My Sleeve” was released to Spotify, seemingly featuring the voices of Drake and The Weeknd. Upon further investigation, it was revealed that the voices heard on the track were AI generated and posted by creator “Ghostwriter”, only to be quickly taken down when Universal Music Group, a rightsholder in Spotify, made a copyright claim citing AI as an unauthorized use of an artist’s work—condemning it as morally and commercially wrong. The removal of the AI work shortly after Spotify had already introduced their own AI features created a brief confusion over what would and would not be allowed on the streaming service, and who would be able to utilize AI technology. Following this, Spotify CEO Daniel Ek clarified that he feels AI can be used to either enhance and tune existing music, or to wrongly infringe on existing artists. Ek’s goal for the platform is to use AI in a way that either legally enhances existing media or be a platform for artists to create AI music inspired by rather than directly copying artists. In furthering this idea, Spotify released their pilot Voice Translation program in which AI is used to translate podcasts into other languages all while using the podcaster’s same voice and vocal characteristics.
The utilization of AI for podcasting prompts new questions about the copyright rights surrounding an individual’s voice. Currently, podcast material best fits under what U.S. law considers a “sound recording.” The Copyright Act defines a sound recording as “works that result from the fixation of a series of musical, spoken, or other sounds but not including sounds accompanying a motion picture or other audiovisual work.” While you cannot outright copyright a voice, the way a voice is used, especially if that of a distinctive or well-known person, could be subject to copyright or other intellectual property laws. For example, actor Morgan Freeman’s distinctive voice itself cannot be protected by copyright, but if a content creator were to use voice recordings and clippings of Freeman from prior work to then be altered by AI, it would likely be considered copyright infringement even if the AI product is different from the original work. Additionally, the Copyright Act provides protection to copyright owners when it comes to any derivative works—new product that recasts, transforms, adapts or is based on an original copyrighted work (for example: a book being translated and published in different languages would be protected by the author’s original copyright). Following the same principles, a podcast and any rendition based off the original material, including a translation would likely be protected by at least these 2 safeguards. However, it is possible to podcast derivatives, and perhaps AI voice translations to be legal under the “fair use” doctrine based on several factors. First, the purpose and character of the podcast use come into play, with transformative and non-commercial uses more likely to be considered “fair.” The nature of the copyrighted work matters, favoring factual over highly creative content as factual information is less susceptible to free interpretation. The amount and substantiality of the material used matter, with smaller portions being more favorable. Finally, the effect on the potential market is crucial; if the podcast use harms the market for the original work, it’s less likely to be considered fair use, with the potential market impact being a key factor in this assessment.
Creators utilizing AI voices may also argue that, under the “originality requirement,” the resulting work must exhibit a “modicum of creativity,” a standard defined by the Supreme Court as even a minimal degree, in the context of copyrightable material. However, under the second prong of the originality requirement, the work in question must be an independent creation of the author. Thus, an AI creation using another person’s existing voice would likely not satisfy the independent creation requirement. For Spotify to effectively expand this program, it will require either existing ownership of the podcast being translated or express authorization from the podcast owner to prevent copyright or other intellectual property issues. To ensure compliance, Spotify and other AI podcasting technologies that do not already own a podcast can apply to receive and pay for a license from the original owner/creator, allowing permission to use the original work. Much like paying for a license to use a song, AI companies would pay for the rights to manipulate and modify a podcaster’s voice for translation into a different language—and thus avoid potential copyright issues.
The ongoing evolution of AI-driven podcasting underscores the critical significance of upholding copyright and intellectual property rights, particularly in the realm of voice copyright. Ensuring compliance through permissions and licensing is essential for the future of platforms like Spotify. By adopting this responsible approach, we can encourage innovation while respecting the rights of content creators, even as we navigate the evolving landscape of voice copyright.
Student Bio: Sarah McLaughlin is a second-year law student at Suffolk University Law School. She is a staff writer on the Journal of High Technology Law and is the current President of the Environmental Law Society. Sarah received a Bachelor of Arts Degree in Political Science and Justice Studies from the University of New Hampshire.
Disclaimer: The views expressed in this blog are the views of the author alone and do not represent the views of JHTL or Suffolk University Law School.
