“Deepfakes” crafted with artificial intelligence are throwing a wrinkle into the murky world of copyright law. JAY-Z and Roc Nation filed a takedown request against YouTube channel Vocal Synthesis for impersonating Hov’s likeness with AI-generated deepfakes.

The creator of the channel trained text-to-speech program Tacotron 2 to mimic Hov’s voice and eventually attracted the attention of the man himself. The account uploaded videos of the computer-generated model reciting Billy Joel’s “We Didn’t Start the Fire” and “To Be or Not To Be” from Shakespeare’s Hamlet.

After receiving the takedown notice, Vocal Synthesis quickly responded with another deepfake featuring Barack Obama and Donald Trump.

AD

AD LOADING...

“Unfortunately, for the first time since the channel began, YouTube took down two of these videos yesterday as a result of a copyright strike,” read Trump’s computerized voice. “The strike was requested by Roc Nation LLC, with the stated reason being that it, quote, ‘unlawfully uses an AI to impersonate our client’s voice.’”

The two videos were initially taken down but eventually reinstated as YouTube investigates.

AD

AD LOADING...

“After reviewing the DMCA takedown requests for the videos in question, we determined that they were incomplete,” a Google spokesperson toldThe Verge. “Pending additional information from the claimant, we have temporarily reinstated the videos.”

Interestingly, other videos featuring Jigga’s likeness remained on the channel while the two in question were removed. The account also features the AI Hov “rapping” verses from the Book of Genesis and a popular Navy Seal copypasta that has circulated the internet for years. The page also uploaded a fifth video where the automated voice raps the “Darth Plagueis” copypasta.

Lil Cease Says Biggie Thought JAY-Z Was A Better MC

The anonymous creator of Hov’s deepfake argued he was protected under Fair Use during an interview with Waxy.

AD

AD LOADING...

“I’m not a lawyer and have not studied intellectual property law,” he said. “But logically I don’t really understand why mimicking a celebrity’s voice using an AI model should be treated differently than someone naturally doing an (extremely accurate) impression of that celebrity’s voice. Especially since all of my videos are clearly labeled as speech synthesis in both the title and description, so there was no attempt to deceive anyone into thinking that these were real recordings of JAY-Z.”

Check out some of the deepfakes above.

[apple_news_ad type=”any”]