OpenAI pulls ChatGPT's 'Sky' voice after users compared it with Scarlett Johansson
OpenAI said in a blog post that the voice was sampled from an undisclosed actor and was not designed to sound like Scarlett Johansson.
Warner Bros. Pictures/Sony Pictures Releasing
- OpenAI plans to pause ChatGPT's "Sky" voice after it was compared with Scarlett Johansson.
- The company said the voice was not an imitation of Johansson and belongs to a different actor.
- Many have compared to the chatbot's voice to Johansson's character in Spike Jonze's "Her."
OpenAI said it would "pause" ChatGPT's "Sky" voice after it sparked comparisons with Scarlett Johansson.
The company shared a blog post on X, saying: "We've heard questions about how we chose the voices in ChatGPT, especially Sky. We are working to pause the use of Sky while we address them."
Many social-media users linked the latest update to Jonze's prophetic film, comparing the voice with Johansson's character in Spike Jonze's 2013 film "Her." The plot features a man falling in love with an AI system.
Some commentators complained that the bot's voice sounded overly sexual and was too flirty in some of the demos.
OpenAI CEO Sam Altman even appeared to address the similarity, posting the word "her" on social media during a recent company demo that heavily featured the voice.
However, the company's CTO, Mira Murati, told The Verge the voice had not been designed to sound like Johansson, adding that someone in the demo's audience had asked the same question.
OpenAI addressed the comparisons with Johansson directly in the blog post, saying the voice was not modeled on Johansson's.
"Sky's voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice. To protect their privacy, we cannot share the names of our voice talents," the company said in the post.
Representatives for OpenAI did not immediately respond to a request for comment from Business Insider, made outside normal working hours.
Last week, the company demoed a new AI model called GPT-4o, which uses native audio inputs and outputs. When integrated into ChatGPT, users can have human-like conversations with the bot, speaking to it and showing it things.
The result is designed to feel like having a real-time virtual assistant, or what could be considered an AI best friend, in your pocket.