OpenAI’s decision to pause Sky, one of ChatGPT’s five voices, became a flashpoint because of its alleged resemblance to Scarlett Johansson. The dispute connects voice casting, celebrity identity, generative AI, and the public expectations around how human-sounding assistants should be built.
The available facts point in two directions at once. Johansson says she was “shocked, angered and in disbelief” by a voice she considered too close to her own. OpenAI says Sky was not an imitation, was performed by another professional actress, and had been selected before the company first contacted Johansson.
What OpenAI says Sky was meant to be
Sky was one of five “carefully selected” voices used to make ChatGPT feel more natural in the smartphone app. OpenAI described its goal as finding a voice that sounded warm, engaging, trustworthy, charismatic, timeless, confidence-inspiring, natural, and rich.
The company emphasized that it did not intend to copy celebrity voices. Its explanation was direct: “Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice.”
OpenAI has not disclosed the actress’s name, citing privacy reasons. According to the source article, the voice was legally licensed, and OpenAI said it worked with renowned casting agencies and directors for five months during the selection process.
The casting process described by OpenAI was broad. More than 400 submissions were reviewed, 14 actors were selected, and OpenAI met with them in person to explain its vision for human-AI interaction. The actors whose voices were chosen traveled to San Francisco for recording sessions in June and July 2023. Their voices were added to ChatGPT in September 2023.
Why Scarlett Johansson objected
Johansson’s concern was not simply that Sky was a synthetic voice. According to NPR, she said the voice sounded very similar to her own and framed the issue as a personal affront and a threat to privacy rights.
“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,” she said in a statement to NPR.
Johansson also said that at a time when generative AI and deepfakes are bringing the protection of one’s image, work, and identity to the forefront, issues like this require “absolute clarity.”
The concern has context in the source article. Johansson was described as one of the most high-profile victims of the first wave of deepfakes, which began in 2017. Her face was deepfaked into numerous porn videos. In late 2023, she sued AI startup Lisa AI for allegedly using her name and image in online advertising without her consent.
That background helps explain why a voice similarity claim could become more than a product-design argument. For a public figure whose identity has already been misused in AI-related contexts, the boundary between inspiration, resemblance, and imitation carries real stakes.
The timeline matters
The central question is whether Sky was created to resemble Johansson after she declined to license her voice. The source article says OpenAI CEO Sam Altman allegedly made several unsuccessful attempts to license Johansson’s voice for ChatGPT, most recently two days before the launch of GPT-4o.
Johansson turned down the deal. OpenAI reportedly licensed a similar-sounding voice, while Altman said Sky had been cast before the first contact with Johansson. He described the pause as a gesture of respect to the actress.
The May 23, 2024 update adds an important detail from the Washington Post. The publication obtained documents from OpenAI and statements from sources indicating that Sky’s voice had been recorded months before Altman contacted Johansson. The same sources said neither Johansson nor the movie “Her” was mentioned in the voice briefings.
The job posting conditions described in that update were specific but not celebrity-based. The actress had to be non-unionized, and the voice had to sound between 25 and 45 years old and be “warm, engaging [and] charismatic.”
The source article also says the actress behind Sky feels personally attacked by the criticism because it concerns her natural voice. She said she had never been compared to Johansson in her private life.
Why “Her” made the dispute louder
The comparison to Johansson did not arise in a vacuum. In the science fiction film Her, Johansson voiced the AI character “Samantha,” with whom the main character falls in love. According to OpenAI, Her is a role model for the advanced AI assistance the company aims to create.
During the introduction of the new AI voice synthesis, Altman wrote “her” on X. That made the public reading of Sky more complicated, even though the later Washington Post details said neither Johansson nor the movie “Her” appeared in the voice briefings.
This is the heart of the controversy: product intent, public symbolism, and individual consent can be interpreted differently. OpenAI’s documented casting timeline supports its claim that Sky was not created as a Johansson imitation. Johansson’s response reflects a broader fear that AI systems can make identity feel reproducible, even when a company says it used another person’s natural voice.
What happens next for ChatGPT voice
OpenAI stopped using Sky, but the broader ChatGPT voice rollout continued to be part of the company’s product direction. The new voice feature was described as coming to paying ChatGPT Plus users in the coming weeks.
According to OpenAI, the feature would let the chatbot handle interruptions smoothly, respond faster, manage group conversations, filter out background noise, and adapt to tone of voice. The source article notes that the current version of the voice was still the old version from the fall of 2023, though it was already running with GPT-4o in the background.
The Sky dispute shows how quickly voice design can become an identity debate. Even when a voice is licensed and performed by another actor, the perception of similarity can raise questions about consent, privacy, and how clearly AI companies should explain their casting choices.
For OpenAI, the available facts in the source article support its position that Sky was not Johansson’s voice and was recorded before Altman contacted her. For Johansson, the issue remains whether a voice that audiences connect to her can be used in a major AI product without creating confusion. That tension is likely to remain central as AI assistants become more conversational, expressive, and human-sounding.