Taylor Swift has filed a trademark application covering her voice and stage image. It includes a photo of her performing in her distinctive bejewelled Eras Tour bodysuit and two voice recordings: “Hey, it’s Taylor” and “Hey, it’s Taylor Swift.”
It’s the latest example of the singer using her status and power to challenge industry norms and assert the rights of artists.
Although the actor Matthew McConaughey beat Swift to the punch by successfully trademarking some of his famous spoken lines of movie dialogue earlier this year, she appears to be the first music artist of note to take the step. The move raises some interesting issues in terms of copyright law and the rights of music artists.
In music, both sound recordings and the songs which they embody are protected by copyright law. Much of the income generated by the music industry is based on the commercial exclusivity to exploit these forms of intellectual property (IP), which that law ensures.
All of Swift’s vocals are protected from copying in terms of being sampled without permission. But the question of whether or not a vocal being performed (or manufactured in the case of AI) to sound like Swift is a copyright infringement is less clear.
In 1988, Bette Midler successfully sued Ford Motor Company for using an impersonator to perform her songs in TV adverts. This case suggests that deliberately copying a singer’s voice, style and tone can amount to passing off. In UK law, passing off is a common law tort involving misrepresentation that causes reputational or financial damage. In the US, similar protection is provided under the Lanham Act, which also guards against misleading imitation.
Swift has also experienced the darker side of deepfakes, including fake pornographic images and AI-generated photos showing her wearing a “Swifties for Trump” T-shirt ahead of the last US presidential election.
While further legislation is being developed in the US to address AI-generated impersonation more broadly, Swift’s move to trademark both her voice and visual likeness suggests a desire to retain stronger legal control over her identity. This may give her a basis for pursuing civil action under common law or intellectual property rights when objectionable AI-generated content appears.
A more complex issue for Swift and artists in general is the use of their existing works as data by AIs to create new works. For example, you could prompt an AI to write a song for me in the style of Fearless-era Swift but performed by a voice that is a cross between Norah Jones and Diana Krall. It would be very hard to prove that any particular piece of IP had been infringed, as the AI would be synthesising dozens of songs and performances to achieve its creation.
However, new content would have been created via the mining of existing IP; a songwriter’s moral right of integrity protects against adaptation of their works without permission, and musicians and songwriters have become increasingly concerned that they are not being recompensed sufficiently, or indeed even consulted about AI using their work.
In the UK, the Musicians Union has launched an initiative demanding consent and remuneration for AI training and AI-generated music.
There have been some positive developments for artists’ earnings, with certain AI companies recently reaching settlements with major music rights holders over the use of training data in music generation. However, artists such as Swift may still need stronger protection to prevent the unauthorised use, imitation and commercial exploitation of their distinctive vocal styles.
(The Conversation via Reuters Connect)