Scarlett Johansson’s lawyers demand transparency from OpenAI over ‘Her’ AI voice
Scarlett Johansson’s lawyers are pressing OpenAI for transparency on how it developed an AI voice assistant, “Sky,” which Johansson claims sounds remarkably like her own voice. This demand follows a series of public comparisons and media coverage, including a segment on Saturday Night Live, noting the similarity.
OpenAI announced that it is halting the use of the Sky voice, stating, “We believe that AI voices should not deliberately mimic a celebrity’s distinctive voice — Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice.” This statement was made in response to the outcry and legal scrutiny following the voice’s public demonstration.
The controversy began after OpenAI CEO Sam Altman, a known fan of the 2013 film “Her,” posted a cryptic message, “her,” on social media after unveiling the new ChatGPT version. This film, which features Johansson voicing a virtual assistant, seemed to inspire the Sky voice, although OpenAI CTO Mira Murati denied any intentional imitation in an interview with The Verge.
Johansson revealed that Altman had approached her nine months ago, proposing that she license her voice for the project, which she declined. Her shock was compounded when OpenAI released a voice that closely resembled hers, leading to her demand for transparency and legislative safeguards.
In a statement to NPR, Altman claimed the voice actor for Sky was chosen before any outreach to Johansson. “Out of respect for Ms. Johansson, we have paused using Sky’s voice in our products,” he said.
This incident highlights the broader implications of AI technology, particularly in protecting personal likenesses and preventing the spread of disinformation. Johansson emphasized the need for clarity and legal protection in an era of deepfakes.
OpenAI has defended its process, noting that Sky’s voice was selected from over 400 casting submissions. The company plans to introduce more voices to match user preferences while maintaining privacy for the voice actors involved. The Sky voice model, along with other advancements in ChatGPT’s new GPT-4o model, was set to launch as a limited “alpha” release for ChatGPT Plus subscribers in the coming weeks.
The case underscores the ethical challenges and societal impacts of advanced AI technologies, with experts warning of potential emotional connections users might form with AI systems, reminiscent of the plot of “Her.” Visar Berisha, an AI speech technology professor at Arizona State University, noted that voice communication’s intimate nature could foster deep user-machine connections, raising concerns about the future interplay between humans and AI.
As AI voice assistants evolve, the industry faces critical questions about protecting individual identities and ensuring ethical development practices, making transparency and regulatory measures increasingly important.
