Instagram does not know what to do about an increasing, AI-powered trend that is sweeping its app, its boss has admitted.
In recent weeks, a growing number of comedians have complained that they have found posts in which people have stolen their routines and put them into the mouths of AI-generated women. In some cases, those posts appear to be aimed at promoting OnlyFans accounts ostensibly run by those AI avatars.
Now, Adam Mosseri, the head of Instagram, said that he is “very worried” about the trend – but that the company is unsure what to do about it.
His comments came in response to a video by comedian George Brett Olson, who shared an example of a post that appeared to have taken his routine word-for-word but used artificial intelligence to give the impression that it had been delivered by an attractive woman. Mr Olson’s routine had been addressing fears over artificial intelligence and the worry that it could disrupt jobs.
Mr Olson noted that the same problem also appeared to be affecting other performing arts. He pointed to examples of creators using fake accounts to take dance videos and present them as being done by an AI avatar.
He is just one of a number of comedians raising awareness of the problem. Comedian Julia Jasunas said that she had been used verbatim by an AI account, saying that “the thing that’s the most concerning is these AI bots are going viral, and people think that they’re real people”.
The account that used Mr Olson’s joke has gathered 300,000 followers, for instance, and links out to an OnlyFans account. The one that took Ms Jasunas’s joke has 250,000 followers.
In a comment posted on Mr Olson’s original video, Mr Mosseri said that it was not practical for Instagram to “try and ban AI content from the platform”. “It’s gonna be harder and harder to identify, and it’s not all bad content (though yes, there is plenty of slop),” he wrote.
“But we are very invested in trying to drive more traffic to original content creators, and if an AI creator is ripping off content wholesale, it would be good for us to work out what we can do about that,” he wrote. “I’m not sure what it would be, but I’ll make sure to talk to the team and look into it.”
In a follow-up comment, posted after Mr Olson shared Mr Mosseri’s response alongside an example of an AI-generated post that appeared to have stolen an artist’s work, the Instagram head admitted that the company did not know what to do about the growing trend. While Instagram users are able to report examples of copyright infringement, dancers, comedians and others have not usually protected their work in that way and so cannot use that route to have posts taken down, he said.
“To be clear, I am very worried about this trend. We are actively working out what to do; it’s just a tricky problem,” Mr Mosseri wrote.
“We already let people make copyright claims against content and take stolen copyrighted material.
“That’s not enough, though, as I assume many comedians and dancers haven’t gone through the trouble to copyright their material. We need a way to help those creatives without letting bad actors make more fake claims (happens all the time unfortunately).”
Read MoreAnthropic and Elon Musk announce share giveaways to win people round to AI
AI giant Anthropic pledges 5% of shares to help society face AI
Anthropic will let outsiders see inside its company amid panic about rogue AI systems
Nasa astronaut feared they would never get home as they flew around the Moon
New US law could completely change how games consoles work
Apple gives major update on the iPhone Duo after launch frenzy
