(San Francisco) A voice that looks exactly like that of Scarlett Johansson: accused of having copied the voice of the actress through ChatGPT, the general director of OpenAI, Sam Altman, apologized Tuesday, immediately announcing the suspension of Sky, the voice generated by artificial intelligence in question.
“Out of respect for Ms. Johansson, we have suspended the use of Sky’s voice in our products. We apologize to Mr.me Johansson for not having communicated better,” said Sam Altman, according to a statement sent by email to AFP by OpenAI.
The actress accuses the company as well as her boss of having voluntarily and without her knowledge copied her voice. “I was shocked, angry and in disbelief that Mr. Altman had developed a voice that sounded so eerily like mine that my closest friends and the media couldn’t tell the difference,” she said. thus explained in a press release, published Monday at the end of the day.
“Last September, I received an offer from Sam Altman, who wanted to hire me to be the voice of the current ChatGPT 4.0 system,” she also said. “He said he thought my voice would comfort people,” she detailed, stressing that she then “declined the offer”.
Scarlett Johansson further explained that Mr. Altman “implied that the similarity was intentional, by tweeting a single word, Her “. A word meaning “she” in English, but also the title of the film where the actress was the voice of an artificial intelligence system and from which the creators of ChatGPT have not hidden their inspiration.
“Sky’s voice is in no way an imitation of that of Scarlett Johansson,” however assured OpenAI in an article posted on its blog, ensuring that it had been developed with the voice of another professional actress.
“ Deepfakes »
Sam Altman reiterated in his statement that “Sky’s voice is not Scarlett Johansson’s and was never intended to sound like her. We distributed the voice of the actress behind Sky before any contact with Mme Johansson.
The actress claimed for her part that she “was forced to hire legal counsel, who wrote two letters to Mr. Altman and OpenAI, […]. Therefore, OpenAI reluctantly agreed to remove the “Sky” voice.
“At a time when we are all grappling with “deepfakes“and the protection of our own image, our own work, our own identity, I think these questions deserve absolute clarity,” she lamented, saying she was waiting “impatiently” for “appropriate legislation to contribute to to guarantee the protection of individual rights.
OpenAI, for its part, detailed how it worked, with professional actors, to create several digital voices, which it called Breeze, Cove, Ember, Juniper and, therefore, Sky.
New voices coming
This case comes a few days after OpenAI announced that it was disbanding its team whose mission was to mitigate the possible long-term dangers of overly intelligent AI. The announcement was marked by the departure of one of the company’s co-founders, Ilya Sutskever, as well as the head of the team, Jan Leike.
“OpenAI must become a company that puts the safety of general AI above all other considerations,” Mr. Leike wrote Friday on We need to do more, that’s our goal.”
The company presented on Monday the new version of its flagship product, with GPT-4o, with improved performance and behavior that is intended to be closer to humans, also making it free for all users.
“In the future, you should expect even more options as we plan to add voices to ChatGPT to better meet diverse user interests and preferences,” OpenAI added on its blog.