Calvin Wankhede / Android Authority
TL;DR
- OpenAI’s ChatGPT-4o demo featured an AI voice eerily much like Scarlett Johansson’s character in a 2013 film.
- The actor launched an announcement as we speak expressing shock and anger over her likeliness within the voice.
- Johansson had obtained a proposal from OpenAI to voice the AI assistant, which she declined.
OpenAI made information final week when it demonstrated its up to date ChatGPT-4o mannequin, which might speak, giggle, and react like a human. Nonetheless, many customers instantly observed that the mannequin reminded them of an *precise* human, Scarlett Johansson.
In a stunning flip of occasions, Scarlett Johansson has now publicly accused OpenAI of utilizing a voice eerily much like hers for certainly one of its ChatGPT-4o voices, “Sky.”
The problem started when final week’s ChatGPT demo used an AI voice that was shortly in comparison with Johansson’s function within the 2013 sci-fi movie Her, during which she voiced an AI private assistant. Including gas to the hearth, OpenAI CEO Sam Altman tweeted the phrase “Her” after the announcement of the brand new ChatGPT model.
OpenAI has since eliminated the “Sky” voice from the mannequin whereas stating that it had developed the AI voice from one other actress whose id stays undisclosed for privateness causes. The corporate maintains that the voice just isn’t an imitation of Johansson however belongs to a special skilled utilizing her pure talking voice.
Within the assertion launched as we speak (by way of NPR), Johansson expressed her shock and anger upon listening to the demo, stating that even her shut pals and information shops couldn’t distinguish the AI voice from her personal.
Nonetheless, Johansson’s assertion reveals a deeper layer of the story. 9 months in the past, Altman allegedly provided her the function of voicing the brand new ChatGPT voice assistant, however she declined as a consequence of private causes. Simply two days earlier than the revealing, Altman reached out once more, urging her to rethink, however the firm went forward with the launch earlier than any settlement might be reached.
In consequence, Johansson has sought authorized counsel and despatched two letters to OpenAI demanding transparency relating to the event strategy of the “Sky” voice. It looks like OpenAI’s takedown of the Sky voice might have been associated to those letters.
Johansson’s assertion highlights issues over deepfakes and the safety of particular person likeness, work, and id within the age of AI. This incident is popping into a significant chapter within the ongoing debate concerning the moral implications of AI know-how.