Last week OpenAI revealed a brand new conversational interface for ChatGPT with an expressive, artificial voice strikingly much like that of the AI assistant performed by Scarlett Johansson within the sci-fi film Her—solely to all of the sudden disable the brand new voice over the weekend.
On Monday, Johansson issued an announcement claiming to have compelled that reversal, after her attorneys demanded OpenAI make clear how the brand new voice was created.
Johansson’s assertion, relayed to WIRED by her publicist, claims that OpenAI CEO Sam Altman requested her final September to supply ChatGPT’s new voice however that she declined. She describes being astounded to see the corporate demo a brand new voice for ChatGPT final week that appeared like her anyway.
“When I heard the discharge demo I used to be shocked, angered, and in disbelief that Mr. Altman would pursue a voice that sounded so eerily much like mine that my closest mates and information shops couldn’t inform the distinction,” the assertion reads. It notes that Altman appeared to encourage the world to attach the demo with Johansson’s efficiency by tweeting out “her,” in reference to the film, on May 13.
Johansson’s assertion says her agent was contacted by Altman two days earlier than final week’s demo asking that she rethink her determination to not work with OpenAI. After seeing the demo, she says she employed authorized counsel to jot down to OpenAI asking for particulars of the way it made the brand new voice.
The assertion claims that this led to OpenAI’s announcement Sunday in a put up on X that it had determined to “pause the usage of Sky,” the corporate’s identify for the artificial voice. The firm additionally posted a weblog put up outlining the method used to create the voice. “Sky’s voice shouldn’t be an imitation of Scarlett Johansson however belongs to a special skilled actress utilizing her personal pure talking voice,” the put up stated.
Sky is one in every of a number of artificial voices that OpenAI gave ChatGPT final September, however finally week’s occasion it displayed a way more lifelike intonation with emotional cues. The demo noticed a model of ChatGPT powered by a brand new AI mannequin known as GPT-4o seem to flirt with an OpenAI engineer in a manner that many viewers discovered harking back to Johansson’s efficiency in Her.
“The voice of Sky shouldn’t be Scarlett Johansson’s, and it was by no means supposed to resemble hers,” Sam Altman stated in an announcement supplied by OpenAI. He claimed the voice actor behind Sky’s voice was employed earlier than the corporate contact Johannsson. “Out of respect for Ms. Johansson, we now have paused utilizing Sky’s voice in our merchandise. We are sorry to Ms. Johansson that we didn’t talk higher.”
The battle with Johansson provides to OpenAI’s current battles with artists, writers, and different creatives. The firm is already defending quite a few lawsuits alleging it inappropriately used copyrighted content material to coach its algorithms, together with fits from The New York Times and authors together with George R.R. Martin.
Generative AI has made it a lot simpler to create real looking artificial voices, creating new alternatives and threats. In January, voters in New Hampshire have been bombarded with robocalls that includes a deepfaked voice message from Joe Biden. In March, OpenAI stated that it had developed a expertise that would clone somebody’s voice from a 15-second clip, however the firm stated it could not launch the expertise due to the way it is likely to be misused.
Updated 5-20-2024, 9 pm EDT: This article has been up to date with remark from OpenAI CEO Sam Altman.
