OpenAI Thought Using a Scarlett Johansson-like Voice in ChatGPT Was a Great Idea. Scarlett Johansson, Not So Much

Although the company has “paused” Sky, the voice in question, the actress is still demanding clarity regarding its creation.

retrato artístico de scarlett johansson
No comments Twitter Flipboard E-mail

By now, most people know that OpenAI has “paused” the voice known as Sky from ChatGPT after accusations that it sounded just like Scarlett Johansson. What not many knew, however, was that in September 2022 OpenAI actually approached Johansson, offering her a role to create a synthetic voice similar to hers.

She declined, but things didn’t end there. At the launch of GPT-4o, the Sky voice the company used for the chatbot sounded so similar to Johansson’s that not even her close friends were able to tell the difference.

Johansson's reaction. The actress claimed to be “shocked, angered and in disbelief that [OpenAI CEO Sam] Altman would pursue a voice that sounded so eerily similar to mine.”

Not a coincidence. Johansson states that OpenAI’s CEO implied that the similarity was deliberate when he referenced Her, the 2013 film in which Johansson voices an AI system very much like the one his company was about to unveil, in a post on X.

Johansson has asked OpenAI for a detailed explanation of how the Sky voice was developed and demands “absolute clarity” amid growing concerns about how easy it is to generate deepfakes of individual voices.

OpenAI’s response. OpenAI removed the voice to protect itself legally. The company has also explained that it works with several professional voice actors to create ChatGPT’s different voices: Breeze, Cove, Juniper, Ember, and Sky.

According to the company, Sky’s voice was based on a professional actress’ natural voice and was never intended to resemble Johansson’s. The actress, however, finds this explanation insufficient and calls for more transparency.

A controversial context. This follows news that OpenAI’s Superalignment team had been disbanded, with former employees accusing it of failing to provide promised resources. This team was in charge of ensuring responsible AI development.

  • Jan Leike–one of those who left the company–declared that OpenAI’s “safety culture and processes have taken a backseat to shiny products.”

This controversy has caught the eye of the public because it involves a celebrity, but it's only one example of the legal and ethical challenges that generative AI developers, especially OpenAI, will face in the future.

Featured Image | Xataka and Midjourney

Related | One of ChatGPT's Voice Options Sounded Like Scarlett Johansson. After Much Debate, the Company 'Paused' It

Home o Index