The YouTube channel Vocal Synthesis creates humorous cultural mashup deepfakes like Bill Clinton reciting “Baby Got Back” lyrics or Jay-Z reading a Hamlet soliloquy. The channel focuses on audio deepfakes where a neural network is trained on recordings and then synthesizes text. Jay-Z’s RocNation filed a copyright claim to get two of these videos removed because the “content unlawfully uses an AI to impersonate our client’s voice,” according to Waxy. Though the move was effective, it’s not legally accurate.
The shield of copyrights — As deepfakes become more impressive and common, the legal system is struggling to keep up. Like in the case of a Kim Kardashian deepfake, the RocNation claim hides behind copyrights and the aggressiveness of a DMCA takedown. In the realm of falsified video and audio, however, this content generally falls under fair use.
Vocal Synthesis transformed Jay-Z’s discography in a humorous way for no commercial benefit and clearly labels all videos as speech synthesis. The resulting audio is a derivative work that bears no resemblance to any of the rapper’s songs. Only two of four Jay-Z videos were pulled from the account and you can access all of the videos on LBRY.
Public figures, especially those in California, have a better legal footing by going after deepfakes for charges like defamation, publicity rights, and other privacy torts. Fortunately for Mr. Knowles and Ms. West, DMCA takedowns are more about punishing first and asking questions later — if at all.
The potential for malicious misuse of deepfakes is vast, so dealing with innocuous cases in the right way should be prioritized before we start setting dangerous copyright precedents.