OpenAI pull Scarlett Johansson's soundalike Sky's voice from ChatGPT

Reading time icon 2 min. read

Readers help support MSpoweruser. We may get a commission if you buy through our links. Tooltip Icon

Read our disclosure page to find out how can you help MSPoweruser sustain the editorial team Read more

Key notes

  • OpenAI removes “Sky” voice from ChatGPT after users say it sounds like Scarlett Johansson.
  • OpenAI says the voice is not an imitation and comes from a different actress.
  • Over 400 voice actors submitted auditions, chosen based on criteria like diversity and warmth.

OpenAI is pausing the use of a voice in ChatGPT after users said it sounded too much like actress Scarlett Johansson. The voice, named Sky, is one of five available options, and the company insists it wasn’t meant to be an imitation.

This comes after the launch of voice capabilities for ChatGPT in September 2023. As per OpenAI, these voices were carefully chosen through a five-month process involving voice actors, casting directors, and industry advisors. 

The voice has started to feel more and more like Scarlett’s voice ever since they have revealed GPT-4o. With all the expressions in the voice now, it feels like one is, in fact, talking with Scarlett Johansson. 

The company maintains that Sky’s voice is not an imitation of Johansson but belongs to a different actress. To protect privacy, they will not disclose the actress’s name.

OpenAI partnered with award-winning casting professionals to define the criteria for the voices. These characteristics included 

  • diversity, 
  • a timeless quality, 
  • approachability, 
  • warmth, and 
  • charisma.

Over 400 voice and screen actors submitted auditions in May 2023. The recordings ranged from answering mindfulness questions to brainstorming travel plans.

After reviewing the submissions, OpenAI discussed its vision for human-AI interactions with the top contenders. The actors were also informed about the technology’s risks and safeguards.

Finally, an internal team selected the final five voices: Breeze, Cove, Ember, Juniper, and Sky. The actors then flew to San Francisco for recording sessions, and the voices were launched in September 2023.

While Sky is being pulled, OpenAI is looking ahead. They plan to introduce a new, more natural Voice Mode for paid users and will include additional voices to cater to diverse preferences.

More here.