Actress was ‘shocked, angered and in disbelief’ when she heard AI voice; company says AI voices shouldn’t mimic a celebrity’s
2 min
Actress Scarlett Johansson criticised OpenAI over a ChatGPT voice she says is “eerily similar” to her own.
The tech company said Monday it was pausing use of the voice, known as Sky, so it could address questions about how it chose the ChatGPT voices. Many people online have drawn comparisons between Sky and Johansson, who voiced an artificial-intelligence assistant in the 2013 sci-fi romance “Her.” The actress said in a statement her closest friends couldn’t tell the difference.
Johansson said OpenAI Chief Executive Sam Altman wanted to hire her last year to provide her voice for ChatGPT’s current system. She declined. When the actress heard Sky, one of five voices the company offers for its AI tool, she said she was “shocked, angered and in disbelief” that Altman would use a voice so similar to hers.
Johansson said her lawyers asked Altman and OpenAI for more details on how they created Sky.
“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.
The voice of Sky was never intended to resemble Johansson, Altman said in a statement Monday evening.
“We cast the voice actor behind Sky’s voice before any outreach to Ms. Johansson,” he said. “Out of respect for Ms. Johansson, we have paused using Sky’s voice in our products.”
In a blog post Sunday, the company said it picked the five voices from more than 400 submissions from actors, looking for voices that sounded timeless and were easy to listen to.
OpenAI said Sky was the natural voice of another actress whom it hired and wasn’t an imitation of Johansson. It wouldn’t name the actress, citing privacy reasons.
The conflict with Johansson adds to the challenges confronting OpenAI, which has been sued by authors, artists and media companies for allegedly using their material without permission or payment. It also serves as a distraction at a time when OpenAI is trying to highlight new products and move beyond its leadership crisis last fall, when the company’s then-board of directors fired Altman for failing to be “consistently candid.” Altman was quickly reinstated as CEO.
OpenAI announced an updated ChatGPT voice feature a week ago. It builds on a product released in September that allows users to talk to its AI tool instead of type and hear responses in five different voices. OpenAI said users can have a more humanlike conversation with the new version, which responds almost instantaneously and can switch quickly between emotional tones.
The updated feature is part of a new AI system , called GPT-4o. It is the company’s latest attempt to attract more users and dominate the market for generative AI technology. The feature will be available to users who pay for ChatGPT-Plus, which costs $20 a month.
At the announcement last week, Altman likened the voice feature to something only seen in movies.
The CEO said in a speech last year that he and other OpenAI executives found inspiration in “Her,” which starred Joaquin Phoenix as a lonely man who falls in love with the voice assistant Samantha, voiced by Johansson. OpenAI employees posted references to the movie on X after the May 13 voice announcement. Altman posted a one-word tweet : “her.”
—Deepa Seetharaman contributed to this article.
Copyright 2020, Dow Jones & Company, Inc. All Rights Reserved Worldwide. LEARN MORE
The Australian leather house has opened an immersive four-day pop-up in Manhattan, unveiling its Bloom Collection and redefining what a product launch can look like.
Following the successful launch of its Palais Collection, MAISON de SABRÉ has unveiled a new modular handbag system offering more than 720 styling combinations.
Meta is betting on a human-first AI future, but growing legal battles and declining public trust are putting Mark Zuckerberg’s vision to the test.
3 min
“Call us optimists. Call us dreamers. Call us whatever the hell you want, but we’re betting on people, and we like those odds. The future is for everyone.”
That ad copy is from the voice-over of a July Meta Platforms META -3.38%.
spot that’s been part of a public-relations blitz to position Meta as the humanist AI company. The message was undercut by the ad’s inclusion of David Bowie’s “Five Years,” a brooding 1972 song about an impending apocalypse. But this week CEO Mark Zuckerberg left no ambiguity, publishing a 6,500-word manifesto—about 10 times the length of this newsletter—with a title that echoed the ad: “The Future is for Everyone.”
That seems to be Meta’s new tagline. In light of sinking public opinion and the company’s thousands of lawsuits from states, school districts, parents, and users, Meta’s public relations have been defensive. This push represents a return to offense, with a chance to distinguish Meta’s approach to AI from other labs like OpenAI, Anthropic, or SpaceX SPCX +9.65%.
“It is surprising that the discourse from many developing AI is so filled with doom,” Zuckerberg wrote. “I do not understand why anyone who believes that AI will eliminate most jobs and much of humanity’s relevance would rush to build that future.”
Zuckerberg frames what sort of future we build with AI as the central issue of our time. “We believe that delivering superintelligence to everyone is the way to answer this question,” he says. “This has the potential to begin a new era of personal empowerment where individuals can use this powerful new capability to reach their full potential, pursue their interests, and improve their lives and the world more than ever before.”
The flood of words belies the situation on the ground in mid-2026. Americans, at least, have a love-hate affair with social media. A November Pew Research Center poll reported that 71% of U.S. adults used Facebook, and 51% used Instagram. Worldwide, 3.6 billion people use at least one Meta app every day.
But in a Reuters/Ipsos poll conducted in July and August, 61% of respondents said they wanted more government oversight of social media, and two-thirds supported laws to keep children under 16 years old off the platforms. When it comes to Meta in particular, in the 2026 Axios Harris 100, an annual poll about corporate reputation, Meta placed 96th out of 100. It’s only above two other social media companies, Chinese ultracheap retailer Temu, and Spirit Airlines, a defunct air carrier. Regarding ethics, Meta came in last, and it was only ahead of TikTok in trust.
The steady drip of headlines in the teen social media trials isn’t helping. Last week, Meta lost a judgment in New Mexico state court that raised their liability in that relatively small jurisdiction to nearly $1 billion dollars. On Wednesday, jury selection began for a federal case with four states suing Meta over addictive product design, and false marketing that said its platforms were safe for teenagers. In July, Meta claimed that the states are asking for a total of $1.4 trillion in damages, in addition to design changes in the apps. This is part of a multidistrict litigation, where thousands of federal trials with social media defendants are coordinated in Judge Yvonne Gonzalez Rogers’ district courthouse in Oakland, Calif.
There is a separate such group of thousands of cases in California state court, mostly with individual plaintiffs. The steady drip of bad headlines from the courts will continue unless Meta decides to settle en masse.
Meanwhile, in the second quarter, Meta booked “$2.40 billion of charges related to legal proceedings,” according to its quarterly filing. That may be just the beginning.
Zuckerberg spent 6,500 words getting his utopian message out, but I can sum it up in two: Trust us. The evidence is that Meta has a long way to go to win back that trust.
