Fighting AI deepfakes (Taylor’s version)

 Fighting AI deepfakes (Taylor’s version)

Andreas Rentz/Getty Images

Taylor Swift has filed three unique trademark applications to protect her voice and image from use in unauthorized deepfakes. It’s a new legal maneuver designed to give celebrities more protections against AI, though it has yet to be tested in court. How would the trademarks defend against unwanted AI slop? Swift moved to trademark two audio clips where she speaks her name, which falls under a lesser-known category called “sound marks.” That has historically been reserved for things like Netflix’s “tu-dum” sound and the giggle of the Pillsbury Doughboy. The other filing is an image of Swift on stage in a sequined outfit and with a pink guitar. The theory is if someone generates AI content with her voice or a similar image, Swift would have more standing in court if she filed a lawsuit against its creators. The bar for violating a trademark is creating something that’s “confusingly similar,” kind of like The Rock’s action movies or The Bachelor contestants. What about copyright? AI creations can be made without lifting a voice from a copyrighted song recording. These new types of spoken-word trademarks, which have also been filed by actor Matthew McConaughey, attempt to plug the gulf that allows deepfakes to flourish.