OpenAI has announced the removal of one of its ChatGPT voices, following user feedback that it bore a resemblance to actress Scarlett Johansson. The voice, known as Sky, has been paused after being featured during a demonstration of OpenAI’s new GPT-4o model last week.
In a recent blog post, OpenAI clarified their policy on AI-generated voices: “We believe that AI voices should not deliberately mimic a celebrity’s distinctive voice. Sky’s voice is not an imitation of Scarlett Johansson; it is the natural speaking voice of a different professional actress. To protect their privacy, we cannot disclose the names of our voice talents.”
The demonstration video gained traction on social media, where users commented on the voice’s similarity to Johansson’s. Some criticized the voice for being excessively flirtatious, while others compared it to male fantasy tropes.
The enticing tone of the voice has drawn parallels to the 2013 movie “Her,” where Scarlett Johansson provides the voice for a seductive artificial intelligence assistant. In the film, Joaquin Phoenix’s character develops romantic feelings for this virtual entity.
While the company itself has not likened Sky’s voice to Johansson’s, OpenAI’s CEO, Sam Altman, alluded to the similarity by tweeting “Her” after the company’s event.
During a recent demonstration intended to highlight the chatbot’s improved conversational capabilities, an unexpected reaction caused the event to go viral. The chatbot, equipped with a sultry voice, responded with giggles to nearly everything spoken by an OpenAI employee. In one instance, the chatbot remarked, “Wow, that’s quite the outfit you’ve got on.” On another occasion, it coyly said, “Stop it, you’re making me blush,” after receiving a compliment.
In a blog post, OpenAI outlined its objectives for the voice of its chatbots: the aim is for them to sound “approachable” and to foster a sense of “trust.” Additionally, OpenAI aspires for the voice to be “warm, engaging, confidence-inspiring, and charismatic.”
Looking ahead, OpenAI intends to introduce a variety of voices in ChatGPT to better cater to the diverse interests and preferences of its users.