Scarlett Johansson is ‘angry’ because OpenAI chatbot imitates ‘her’ voice

Scarlett Johansson criticized OpenAI and founder Sam Altman on Monday, saying the artificial intelligence company, which makes ChatGPT, created a new chatbot voice that sounded “eerily similar” to hers, after she refused to grant permission. License your voice to the system.

“When I heard the published demo, I was shocked, angry, and incredulous that Mr. Altman would pursue a voice that sounded so eerily similar to mine,” Johansson wrote in a statement published by NPR on Monday.

AI Atlas Art Badge Sticker AI Atlas Art Badge Sticker

OpenAI last week showed off GPT-4 Omni (GPT-4o), its latest AI model. The company demonstrated how AI can converse in a more human way, with the ability to whisper, make sarcastic comments and even flirt.

OpenAI’s virtual assistant demo quickly drew comparisons to Johansson’s character from the 2013 film Her. In that film, directed by Spike Jonze, Johansson played Samantha, a virtual assistant who develops an intimate relationship with a reclusive writer.

Read more: Microsoft Copilot embraces the power of OpenAI’s GPT-4o

See this: Scarlett Johansson says ChatGPT sounds like her; OpenAI eliminates voice

On Sunday, OpenAI said it has suspended use of its AI voice, called Sky, while it addresses issues surrounding the issue of its virtual assistant’s voice. Sky, which has been available since OpenAI launched the ChatGPT voice mode last September, was one of five voices available with GPT-4o. In a blog post published on Sunday, OpenAI said it did not copy 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 rather belongs to a different professional actress using her own natural voice,” the company said. “To protect her privacy, we cannot share the names of our announcers.”

A large screen on stage showing the words. "Project Astra" A large screen on stage showing the words.

Google introduced its own real-time multimodal AI assistant shortly after OpenAI introduced GPT-4o.

Numi Prasarn/CNET

Altman reiterated the company’s stance on Monday. The former CEO of OpenAI said that Sky’s voice was “never intended to resemble” Johansson’s voice, in a statement shared with CNET.

“We chose the voice actor behind Sky’s voice before contacting Ms. Johansson. Out of respect for Ms. Johansson, we have stopped using Sky’s voice in our products,” Altman said. “We are very sorry, Ms. Johansson, that we have not communicated better.”

Read more: The best phone to buy in 2024

In her statement, Johansson accused the company and altman to intentionally copy his voice. The American actor said that Altman had approached her in September about giving voice to GPT-4o to help consumers “get comfortable with the radical change affecting humans and AI”, adding that Altman said that his voice would be “comforting for people.” Johansson, who rejected the initial offer for personal reasons, said Altman had approached his agent again days before the May event, asking her to reconsider licensing his voice for a virtual assistant.

Screenshot of an X post by Sam Altman with just one word: "his" Screenshot of an X post by Sam Altman with just one word:

Sam Altman posted the word “she” in X in connection with OpenAI’s GPT-4o event.

X/CNET Screenshot

“Before we could connect, the system was already available,” Johansson said in his statement. She added that she was “forced” to hire legal counsel, who wrote to Altman asking for transparency about the process followed to hire the host.

When choosing an AI voice, according to OpenAI’s blog post, the company narrowed down more than 400 voice and screen actor submissions to just five, whose voices it believed embodied a set of characteristics including “an approachable voice that inspires confidence.” “, “that feels timeless.” and is “natural and easy to listen to.”

The legal threat posed by Johansson comes as OpenAI faces a series of copyright violations by creative industries spanning Hollywood and the broader media industry. In April, a group of eight newspapers took legal action against OpenAI and Microsoft, filing a lawsuit alleging copyright infringement related to the unauthorized use of their articles to train AI models.

More from Google I/O 2024

Editor’s note: CNET used an artificial intelligence engine to help create several dozen stories, which are tagged accordingly. The note you are reading is attached to articles that substantially cover the topic of AI, but were created entirely by our expert editors and writers. For more information, see our AI Policy.