Copyright Claims Against Deepfaked Audio Raises Legal Questions


Vocal Synthesis, a YouTube channel featuring deepfaked audio, allegedly received a copyright claim and took down two videos of Jay-Z impersonations.’Deepfakes’ are sophisticated imitations created through the use of artificial intelligence. According to the anonymous content creator behind Vocal Synthesis, the copyright claims were filed by Roc Nation LLC, whose reason for removal states, “This content unlawfully uses an AI to impersonate our client’s voice.” Although the Jay-Z videos have since been restored, the copyright claim raises more questions about deepfake legality.

Audio deepfakes are AI-generated speech mimicking the human voice. The mechanism for its creation includes deep learning and training generative neural network architectures with an expansive audio collection. This technology is controversial because it could become a tool for deception and the spread of disinformation. However, Vocal Synthesis intends to entertain rather than deceive. The description for each video on the channel says, “The voice in this video was entirely computer-generated using a text-to-speech model.”

The creator behind Vocal Synthesis explained his own perspective on the issue in a response video featuring the voices of Presidents Donald Trump, Barack Obama, and JFK. “The channel was created by an individual hobbyist with a huge amount of free time on his hands, as well as an interest in machine learning and artificial intelligence technologies,” said the voices. “He would like to emphasize that all of the videos on this channel were intended as entertainment, and there was no malicious purpose for any of them.”

There have been some legislative responses to deepfakes in the United States. The Senate proposed the Malicious Deep Fake Prohibition Act of 2018, and the House of Representatives introduced the Defending Each and Every Person from False Appearances by Keeping Exploitation Subject to Accountability (DEEPFAKES) Act of 2019. Facebook banned deepfakes from its platform in 2018, but left open the possibility of conventionally-created “shallow fakes.” California banned deepfakes in 2018, although only for videos in a political context.