OpenAI has been forced into a rapid about turn after Scarlett Johansson accused the firm of using “a voice that sounded so eerily similar to mine” for its latest release.
The actor has called for “legislation” to protect individuals’ rights as AI firms race to digitise us and everything we’ve ever produced, then charge us for the privilege.
ChatGPT-4o’s ersatz Scarlett, under the name Sky, was unleashed on the world just weeks after OpenAI had unsuccessfully asked Johansson to lend them her voice. She had previously knocked back the firm in 2023.
Further complicating matters is that Sam Altman had expressed his enthusiasm for the film Her, where Joaquin Phoenix’s character falls in love with a Johansson voiced AI assistant.
When GPT-4o was launched earlier this month, observers rhapsodised about the apparent flintiness of the Sky voice. However, others drew a direct comparison with Johansson’s voice, as well as suggesting that a “flirtatious” bot represented a “mega ick”. Even Elon Musk admitted to a “cringe” on hearing the demo.
OpenAI explained in a post over the weekend that it had worked with a variety of voice actors to develop the voices, paying them “above top-of-market rates”. Interestingly, it implied this was an ongoing payment “for as long as their voices are used in our products”, rather than a one-off fee.
It insisted that “Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice.”
Yesterday, Johansson went public. In a statement to NPR tech correspondent, Bobby Allyn, she said she had declined an offer from Altman for her voice last year, not least as it could “bridge the gap between tech companies and creatives”. She added that the OpenAI boss had said “He felt my voice would be comforting to people.” She declined. Again.
See also: OpenAI’s “Sora” lets you create video from text: Be awed, be worried.
She also said that Altman contacted her agent two days before the release of 4o “asking me to reconsider. Before we could connect, the system was out there.”
Johansson said she was “forced to hire legal counsel” who set out the situation, and asked Altman and OpenAI “to detail the exact process by which they created the ‘Sky’ voice.”
Instead, it seems, “OpenAI reluctantly agreed to take down the Sky voice.”
Johansson’s statement noted that “we are all grappling with deepfakes and the protection of our likeness, our own work, our own identities.” These questions deserve “absolute clarity” she said, and “appropriate legislation to help ensure that individual rights are protected.”
While Johansson has scored a victory, creators in general are grappling with how technology changes, well, everything.
Last year’s Hollywood actors’ strikes were in large part about the film industry’s eagerness to pay peanuts to replicate actors in AI.
The New York Times filed a suit in December accusing OpenAI of copyright infringement, and that case is still ongoing.
Meanwhile the FT has struck a deal with OpenAI around content, while Reddit has struck agreements with both OpenAI and Google.
It’s fair to say that creators’ rights around their likenesses and content are contested. However, Johansson might just have moved the dial slightly.
Demonstrating just how up to date ChatGPT is, when we asked it “why is Scarlett Johansson unhappy today?”, it noted the imbroglio, adding “This incident has brought attention to the broader issues of AI ethics and the need for stricter regulations regarding the use of personal likenesses in AI technologies.”
It also kindly flagged up her latest projects, adding that they and her stance on AI were “currently contributing to her pres
