Scarlett Johansson released a statement regarding the OpenAI situation pic.twitter.com/VuSQhkouVt
— Best of Scarlett Johansson (@BestfScarlett) May 20, 2024
Altman reached out to the actor about the project last September, which she declined to be involved with, but it wasn't until this month that her friends, family and the public all remarked how much the "Sky" system sounded like her, Johansson said in her statement.
"When I heard the released demo, I was shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference," Johansson said Monday.
Altman even appeared to acknowledge the similarity to Johansson's voice when on May 13, he posted on X the single word "her," seemingly in reference to the 2013 Spike Jonze film of the same name, which follows a character played by Joaquin Phoenix who develops an intimate relationship with an AI virtual assistant personified through a feminine voice played by Johansson.
her
— Sam Altman (@sama) May 13, 2024
The actress said she was forced to hire attorneys, who wrote letters to Altman and OpenAI asking them to detail the process of creating the "Sky" voice.
"In a time when we are all grappling with deepfakes and the protection of our own likeness, our own work, our own identities, I believe these are questions that deserve absolute clarity," Johansson said. "I look forward to resolution in the form of transparency and the passage of appropriate legislation to help ensure that individual rights are protected."
In a statement to Law360 Monday evening, Altman said, "The voice of Sky is not Scarlett Johansson's, and it was never intended to resemble hers. We cast the voice actor behind Sky's voice before any outreach to Ms. Johansson."
"Out of respect for Ms. Johansson, we have paused using Sky's voice in our products," he added. "We are sorry to Ms. Johansson that we didn't communicate better."
Late Sunday, OpenAI took down "Sky" and published a blog post with more information about how its ChatGPT voices — which also include "Breeze," "Cove," "Ember" and "Juniper" — were created. The company said it worked with voice actors, paying them "above top-of-market rates."
OpenAI specifically addressed the issue of mimicking celebrity voices, saying the "Sky" voice "is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice," according to the blog post.
Of the more than 400 submissions it received from voice and screen actors, Open AI said it chose the five voices it is currently using. But to protect the actors' privacy, the company declined to share their names.
We've heard questions about how we chose the voices in ChatGPT, especially Sky. We are working to pause the use of Sky while we address them.
— OpenAI (@OpenAI) May 20, 2024
Read more about how we chose these voices: https://t.co/R8wwZjU36L
"We believe that any broad deployment of synthetic voice technology should be accompanied by voice authentication experiences that verify that the original speaker is knowingly adding their voice to the service and a no-go voice list that detects and prevents the creation of voices that are too similar to prominent figures," the company said in a March post.
Johansson is not the first actor to wonder if her voice has been stolen for artificial intelligence.
Two voice actors last week accused AI startup Lovo of using their voices without permission and without properly compensating them, after they said the company tricked them into providing voice samples under the guise of freelance narration work.
"Lovo represents to potential clients that it has agreements with actors allowing Lovo to utilize those actors' voices and compensating them appropriately for that use," the voice actors say.
"That may be true with respect to some actors, but it is emphatically not true with respect to plaintiffs: Lovo had no permission to use plaintiffs' voices for training its AI generator, to promote the Lovo service, or to market voices based on plaintiffs' voices," they add. "And Lovo never compensated plaintiffs for any of Lovo's unauthorized uses of plaintiffs' voices."
And this isn't the only flak OpenAI has received for its artificial intelligence technology.
Late last month, eight regional newspapers owned by private equity giant Alden Global Capital sued OpenAI and Microsoft Corp., accusing the tech companies of ripping off the newspapers' copyrights and misappropriating news articles to train AI chatbots that also allegedly spread fake news falsely attributed to the newspapers.
Also last month, OpenAI asked to toss a lawsuit from two alternative news websites asserting copyright infringement, as well as a similar one from The Intercept.
--Additional reporting by Dorothy Atkins, Andrew Karpan and Ivan Moreno. Editing by Kristen Becker.
For a reprint of this article, please contact reprints@law360.com.