Did you see the story this past week about AI-written parody song lyrics were DMCA-ed from Twitter? I for one don’t think it’s been reported accurately.
Georgia Tech researcher Mark Riedl didn’t expect that his machine learning model “Weird A.I. Yancovic,” which generates new rhyming lyrics for existing songs would cause any trouble. But it did.
On May 15, Reidl posted an AI-generated lyric video featuring the instrumental to Michael Jackson’s “Beat It.” It was taken down on July 14, Reidl tweeted, after Twitter received a Digital Millennium Copyright Act takedown notice for copyright infringement from the International Federation of the Phonographic Industry, which represents major and independent record companies.
“I am fairly convinced that my videos fall under fair use,” Riedl told Motherboard of his AI creation, which is obviously inspired by Weird Al’s parodies. Riedl said his other AI-generated lyric videos posted to Twitter have not been taken down.
Riedl has contested the takedown with Twitter but has not received a response. Twitter also did not respond to Motherboard’s request for comment.
My problem with this story is that it is being framed as the industry taking action, when for all we know this could just be a copyright bot taking action on its own. No, I do not have anything to base that on, but it is about ten times more likely.
Only the one video was removed when they are all equally legal or illegal. If the _industry_ is going after these videos, then why did they not file DMCA takedown notices for all the other videos?
Why was there no press release?
Why was there no letter to the university, or to the college professor who created the videos?
The most likely explanation is a copyright bot found one video, jumped to the wrong conclusion, and filed the DMCA notice all on its own. We all know that automated systems do this all the time; in fact, just this past week Youtube’s ContentID shutdown two different livestreams from San Diego Comic-con. And previously, Digimarc showed us in 2016 what could happen if you let a bot file hundreds of (bogus) DMCA notices without supervision.
So until someone can show me that a person was involved in the decision loop somewhere, I am going to conclude this is an example of a bot-on-bot legal fight.
That just makes more sense to me.