Scarlett Johansson says that OpenAI asked her to be the voice behind ChatGPT — but that when she declined, the company went ahead and created a voice that sounded just like her

OpenAI co-founder Sam Altman is facing an unexpected adversary, Hollywood actress Scarlett Johansson, who has become embroiled in a dispute over the voice used for the ChatGPT text-to-speech product.

According to reports from the New York Times, Johansson revealed that Altman had approached her in September to lend her voice to the product, citing her voice as "comforting." However, she declined the offer. Despite this, when ChatGPT unveiled its voice feature nine months later, one of the options bore a striking resemblance to Johansson's voice, leaving her "shocked, angered, and disbelieving" at the apparent similarity.

Johansson disclosed that she sought legal counsel to inquire about how OpenAI had developed the voice, named Sky. Altman, a known fan of the movie "Her" where Johansson voiced a character, had previously expressed admiration for her work.

OpenAI's voice feature, introduced in 2023, allows users to interact with ChatGPT through spoken commands and offers five voice options for text-to-speech responses.

Despite Altman's assertion that Sky's voice was not modeled after Johansson's, OpenAI paused its use in response to user queries regarding the voice's selection and sampling process.

In a statement to CBS MoneyWatch, Altman clarified that the voice actor for Sky was chosen before any contact with Johansson, expressing regret for the misunderstanding.

OpenAI announced the temporary suspension of Sky's voice through X (formerly Twitter), aiming to address concerns raised by Johansson and others.

In a subsequent blog post, OpenAI emphasized that its voice selection process prioritizes originality and professionalism, denying that Sky's voice was an imitation of Johansson's.

The company explained that it collaborated with casting professionals and voice actors to narrow down voice options from a pool of 400, selecting winners based on various criteria, including timeless appeal and ease of listening.

Johansson's representatives stated that Altman had approached her twice regarding the voice project, which she declined. However, the voice was unveiled before any further discussions could take place.

Upon hearing the released demo, Johansson expressed her dismay at the perceived similarity to her voice and initiated legal action to understand the voice creation process. Only after these actions did OpenAI remove the voice from its product lineup.