In Short:
OpenAI introduced a new ChatGPT voice similar to Scarlett Johansson’s, prompting her to demand clarity on how it was made. Johansson declined to work with OpenAI last September but the company still developed a voice like hers. After legal action, OpenAI paused the voice. This incident reflects OpenAI’s issues with creatives and the potential misuse of AI-generated voices, including deepfakes and voice cloning technology.
Last week, OpenAI unveiled a new conversational interface for ChatGPT featuring a synthetic voice that closely resembled the AI assistant depicted by Scarlett Johansson in the movie Her. However, the company abruptly disabled the new voice over the weekend.
Statement from Scarlett Johansson
On Monday, Scarlett Johansson issued a statement asserting her involvement in the reversal of the decision after her legal team demanded clarity from OpenAI regarding the creation of the new voice.
In her statement, Johansson revealed that OpenAI CEO Sam Altman had requested her to provide the voice for ChatGPT last September, but she declined. To her surprise, the company presented a new voice for ChatGPT last week that closely resembled hers.
Johansson expressed shock, anger, and disbelief at the striking similarity between the new voice and hers. She cited a tweet by Altman referring to the movie Her in connection with the demo.
Response from OpenAI
In response, OpenAI announced on X that it would “pause the use of Sky,” the name given to the synthetic voice that resembled Scarlett Johansson’s. Although Sky was introduced last September, its more lifelike intonation and emotional cues in the recent demo drew parallels to Johansson’s performance in Her.
When questioned about the decision to disable Sky, OpenAI spokesperson Niko Felix directed inquiries to a blog post explaining the voice selection process. The post clarified that Sky’s voice belonged to a different professional actress and was not an imitation of Scarlett Johansson.
Implications and Challenges
This incident contributes to OpenAI‘s ongoing legal battles with artists, writers, and creatives over alleged unauthorized use of copyrighted material to train its algorithms. The company faces lawsuits from entities such as The New York Times and authors like George R.R. Martin.
Advancements in generative AI have facilitated the creation of realistic synthetic voices, presenting both new opportunities and risks. Instances like deepfake robocalls featuring political figures and voice cloning technologies highlight the potential misuse of such innovations.
