A live conversation with ChatGPT has sparked a dispute between OpenAI and Scarlett Johansson. She sees too much similarity with her own voice.
In brief:
- ChatGPT-4o can conduct real-time conversations with users.
- Scarlett Johansson finds that one of the voices used sounds very similar to hers.
- She did not permit OpenAI to use her voice for the AI model.
Hollywood star Scarlett Johansson is demanding an explanation from ChatGPT creator OpenAI as to why a voice presented by the company sounds very similar to her own. Her lawyers have sent two letters to OpenAI and the company’s CEO, Sam Altman. Johansson shared this information in a statement published by NPR.
ChatGPT-4o sounds like Scarlett Johansson
The voice recently heard in a demonstration of ChatGPT was “eerily similar” to hers. Even her closest friends and journalists could not tell the difference, the actress emphasized. OpenAI countered that the basis for the AI voice was sentences spoken by another actress. It was never intended to sound like Johansson.
The now-public dispute explains why OpenAI announced that the voice named “Sky” would be temporarily discontinued. Johansson described that Altman had approached her as early as September of the previous year. He had offered to use her distinctive voice as the basis for the ChatGPT AI voice. After careful consideration, she declined for personal reasons.
AI like “from the movies”
The offer has a backstory: In 2013, Johansson voiced an AI software character in the film “Her,” with whom the protagonist falls in love.
After a demonstration of ChatGPT’s new capabilities, Altman himself made the comparison: He wrote on X that the software was like AI from movies – and included the word “her” in his message.
Johansson “shocked and angry”
Johansson wrote that she was “shocked and angry” when she heard the AI voice from the OpenAI demonstration. Clarifying the circumstances is important. Especially in times when society is confronted with deepfakes, transparency and effective laws are needed. It’s about “protecting our likeness, our own work, our own identities.”
“Sky” was introduced alongside four other voices – “Breeze,” “Cove,” “Ember,” and “Juniper” – in September 2023 with ChatGPT. It is only with the new AI model GPT-4o that the chatbot can conduct truly fluid conversations. The five voices are based on sentences spoken by people for OpenAI.
They were selected from over 400 applications from film and stage actors, according to a company blog post. OpenAI does not disclose their names – “to protect their privacy,” as stated.
