This week, Academy Award–nominated actress Scarlett Johansson accused OpenAI, the company behind ChatGPT, of stealing her voice to create an AI virtual assistant called “Sky.” According to Johansson, OpenAI CEO Sam Altman approached her nine months ago about licensing her voice to the company. She declined. Then, two days before the company launched the new version of ChatGPT, he asked again, but before she could respond, the company unveiled its new product, complete with a voice Johansson says is “so eerily similar” to her own. After Johansson complained, OpenAI took down “Sky.”
For an argument about the use of cutting-edge technology, the Johansson-Altman bust-up evokes a case from the 1980s, says Taylor Barkley, director of public policy at the Abundance Institute, a think tank that specializes in AI and other emerging technologies. In the 1980s, Ford wanted to use Bette Midler’s voice in an advertisement. When she declined, they hired a Midler impersonator. “The court decided in Midler’s favor, given the facts leading up to the use of likeness.”
In other words, intent is crucial. And while Open AI denies it sought to imitate Johansson, Altman’s case isn’t helped by the fact that, on May 13, just a few days before the new product’s release, he cryptically tweeted the word Her—in an apparent reference to the 2013 film starring Johansson as an AI chatbot who falls in love with Joaquin Phoenix. (Most normal people see the film as a lonely, dystopian comment on a world fueled by AI, but our tech overlords apparently feel differently.)
“The fact pattern of OpenAI approaching Scarlett Johansson, her declining, you know, at least twice. And then there’s that tweet that Sam Altman still has up. I think those things are what could tip the scales in Scarlett Johansson’s favor,” says Barkley. Not that anyone is suing anyone just yet.
But can old rules keep up with AI technology developing at breakneck speed? In a world beset by deep fakes, it’s an important question that matters to ordinary people as well as rich and famous actors like Johansson.
A number of states have passed legislation banning deep fake porn specifically. And Congress is currently considering the No AI Fraud Act, which, as Barkley explains, “says in general that every individual has a property right in their own likeness and voice.” Altman and his colleagues at OpenAI are pressing ahead with new technology. As for the rest of us, Barkley says: “We’re all figuring this out together, I think culturally and from a policy and legal standpoint.”
River Page is a reporter for The Free Press. Follow him on X @river_is_nice and read his piece, “The Best Buyer for TikTok.”
To support The Free Press, subscribe today:
Well, apparently this article answers the query in the title: "No, she can't".
And really, who cares? AI is a threat on so many deeper levels that articles like this are the social equivalent to the Law of Triviality, otherwise known as the bike-shed problem.
I don't give a shit where Scarlett Johansson puts her goddamned bicycle... AI is more dangerous than any of this...
Who is Scarlett and why would we care what she thinks?...her job is to fake being someone else in films that actually are not real...more interested in how AI can create killing machines than I am about an actress"s voice being faked.....