OpenAI has suspended a voice option in its ChatGPT product that closely resembled Scarlett Johansson’s voice, following the actor’s objections. Johansson expressed her surprise and anger over the use of a voice that was eerily similar to her own without her consent.
Johansson revealed in a statement that OpenAI had approached her nine months prior to voice its AI system, but she declined for personal reasons. Despite her refusal, she was “shocked” to hear the voice option, named “Sky,” which was indistinguishable from her voice according to friends and news outlets.
In response to the backlash, OpenAI removed the heavily promoted Sky voice option from ChatGPT on Monday. The company had showcased Sky during a high-profile event last week, demonstrating the capabilities of the new ChatGPT-4o model. The voice immediately drew comparisons to Johansson’s AI companion character in the 2013 film “Her.”
Adding to the controversy, OpenAI’s CEO Sam Altman had posted a tweet saying “her” following the presentation, implying an intentional resemblance to Johansson’s character. However, OpenAI later clarified that Sky was not based on Johansson, stating in a blog post that the voice belonged to a different professional actress using her natural speaking voice. The company chose not to disclose the names of their voice talents to protect their privacy.
Johansson claimed that Altman had initially pitched the project to her, suggesting that her voice could help bridge the gap between technology companies and creatives and make consumers more comfortable with AI. He reached out again through her agent just two days before Sky was released, asking her to reconsider. Johansson’s lawyers subsequently contacted OpenAI to demand the removal of the voice.
Criticism of Sky was not limited to its similarity to Johansson. Some commentators, including The Daily Show host Desi Lydic, criticized the voice for being overly fawning and gendered, with Lydic joking that it was evident “a man built this tech.”
OpenAI defended its selection process for ChatGPT’s voice, stating it aimed for a voice with a “timeless” quality that would inspire trust. The company reviewed hundreds of submissions before selecting five voice options for ChatGPT, which were released in September. The chosen actors then participated in recording sessions to train OpenAI’s models.
The removal of Sky coincided with significant internal turmoil at OpenAI, with several top members of its safety team resigning. One key researcher, Jan Leike, criticized the company for prioritizing “shiny products” over safety protocols. Despite this, Altman and co-founder Greg Brockman insisted that OpenAI would not release a product if there were safety concerns.
OpenAI’s blog post emphasized its collaboration with entertainment industry professionals and compensation for voice actors. The tech industry, including OpenAI, faces intense scrutiny and legal challenges from entertainers, creators, and media companies over potential copyright violations and fears that AI could replace human workers. Major entertainment unions, such as Sag-Aftra, have gone on strike over issues including the use of likenesses by AI.
