Hollywood actress Scarlett Johansson, on Monday, alleged that OpenAI developed a voice for the ChatGPT system that bore a striking resemblance to her own voice, following her refusal to lend her voice to the chatbot.
Johansson's remarks came in a statement issued shortly after the artificial intelligence company announced the removal of the voice, dubbed 'Sky'. She said that she was left "shocked" and "angered".
OpenAI clarified that it would remove the voice, but insisted that it was not meant to be an "imitation" of the star. The company further stated that they are taking down the AI voice “out of respect for Ms Johansson”.
Johansson accuses OpenAI of unauthorised use of her voice
In a personal claim, Johansson stated that OpenAI initially approached her to use her voice in their novel project. OpenAI, the creator of ChatGPT, unveiled a novel AI model named GPT-4o, proficient in engaging in lifelike voice conversations and interacting seamlessly with both text and images.
During a livestream event, OpenAI researchers had demonstrated the model's new audio features, allowing users to engage in real-time dialogue with ChatGPT without delays and even interrupt it while speaking—challenges traditionally encountered by AI voice assistants.
Voice mode in ChatGPT and the Sky voice model have existed since the previous year. However, its prominence escalated last week as OpenAI showcased enhancements integrated into its latest GPT-4o model. The upgraded model rendered the voice assistant more emotive, enabling it to interpret facial expressions via a phone's camera and provide real-time spoken language translation.
Post the release, many commentators were quick to draw comparisons between the chatbot's tone and Johansson's in the 2013 film 'Her'. Johansson alleged that the company, along with its founder Sam Altman, intentionally replicated her voice. The actress stated that she had declined a prior request from the company to provide the voice for its new chatbot.
Last week, in an interview with The Verge, OpenAI CTO Mira Murati refuted any suggestion of intentional imitation of Johansson. OpenAI's CEO Sam Altman also said in a statement, “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.”
OpenAI added that the five ChatGPT voice profiles presently accessible were chosen from a pool of over 400 casting submissions, including both voice and screen actors. However, the company opted not to disclose the actors' identities, citing the necessity to "protect their privacy".
Referring to her legal recourse, the actress stated, “As a result of their actions, I was forced to hire legal counsel, who wrote two letters to Mr Altman and OpenAI, setting out what they had done and asking them to detail the exact process by which they created the ‘Sky’ voice. Consequently, OpenAl reluctantly agreed to take down the 'Sky' voice."
Johansson's statement further read, “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."
Copyright challenges in the era of AI
The controversy follows just six months after actors ended strikes that had halted the entertainment industry, during which they demanded better pay and protections against the use of AI.
The Hollywood strike, ignited by actors and writers in 2023, had brought significant attention to the impact of AI on the entertainment industry. At the heart of the dispute was the fear that AI technology could replace human creativity, with actors and writers concerned about the unauthorised use of their likenesses and scripts. They argued that studios could potentially use AI to generate performances and write scripts, undermining their labour and intellectual property rights.
Johansson had also participated in last year's strike, which ended in agreement with the studios and included guarantees that such technology would not be employed without the actors' consent.
OpenAI has been facing several legal challenges regarding its use of copyrighted online content. In December, the New York Times announced its intention to sue the company, alleging that it had used "millions" of articles published by the media organisation to train its ChatGPT AI model.
Additionally, in September, authors George R.R. Martin and John Grisham revealed plans to file a claim, alleging that their copyrighted works were used without permission to train the system.
The comedian Sarah Silverman was amongst the first to sue the company on comparable grounds, in July last year. She had joined a class-action lawsuit against OpenAI accusing the company of copyright infringement, saying they “copied and ingested” her protected work in order to train its AI models.
The lawsuits, in which she joined authors Christopher Golden and Richard Kadrey, alleged that OpenAI copied their works, including Silverman’s memoir, "The Bedwetter", without permission. The accusations claimed the company scraped illegal online “shadow libraries” that host the texts of thousands of books.