OpenAI 'pauses' one of ChatGPT's voices for sounding too much like Scarlett Johansson's

OpenAI a week ago it presented its latest language model, GPT-4o, that now powers ChatGPT. During the event he attracted attention improved voice capabilities of artificial intelligence, with response times similar to those of a conversation between humans. There were several examples of its use, interacting with the AI ​​from a smartphone that, inevitably, remind us of the AI ​​in the movie Her by Spike Jonze, with whom Joaquin Phoenix's character fell in love and which the actress voiced Scarlett Johansson. The company has announced today that it is 'pausing' Sky, one of the 5 voices that ChatGPT can use, after the wave of comments and criticism received due to its resemblance to that of the actress.

Sky is not something new on ChatGPT. Voice has been integrated with AI since last Septemberbut it has been the commotion caused by ChatGPT's new real-time chat capabilities have fueled interest. It is also more expressive and can interpret the user's facial expressions through the mobile camera.

'We've heard questions about how we choose voices on ChatGPT, especially Sky. We are working to pause Sky usage while we address them', explains the company in X.

Prior to this message, OpenAI has published a statement on its blog explaining How the voice selection process for AI is carried out, in which he specifically comments on the case of Sky and her resemblance to Scarlett Johansson's voice. 'We believe that AI voices should not deliberately imitate a celebrity's distinctive voice: Sky's voice is not an imitation of Scarlett Johansson, but belongs to a different professional actress who uses her own natural voice. To protect her privacy, we cannot share the names of our speakers,' the company states. OpenAI has not indicated If the possibility of a lawsuit by the actress has to do with her decision.

The 5 voice profiles – Breeze, Cove, Ember, Juniper and Sky – come from voice actors who were collaboratively selected'with casting directors and independent producers'. OpenAI received more than 400 proposals for voice and film actors among whom he was looking for 'actors of diverse origins or who can speak several languages. A voice that feels timeless. An accessible voice that inspires confidence. A warm, attractive, charismatic and confidence-inspiring voice with a rich tone. Natural and easy to listen to'. The process was carried out in early 2023 and the 400 proposals were first reduced to 14 and then to those finally chosen. The recordings of their voices were carried out last summer.

The rollout of the new ChatGPT capabilities is gradual and the new ChatGPT voice assistant capabilities They will be released 'in the coming weeks' as an 'alpha' version -software in an early stage of development- for subscribers of ChatGPT Plus. OpenAI plans to later introduce additional voices to 'better suit users' diverse interests and preferences.'