OpenAI is currently in discussions with Scarlett Johansson's representatives following the removal of a ChatGPT voice that users found strikingly similar to the actress. This move comes after Johansson's legal team demanded transparency regarding the development process of the voice, named Sky.
Legal Demands and Public Reactions
Lawyers representing Johansson have sent two letters to OpenAI, seeking detailed information on how the AI assistant's voice was created. The actress's team has expressed concerns about the voice's resemblance to her character from the 2013 film “Her,” which has prompted a broader conversation about the ethical implications of AI-generated voices.
Joanne Jang, who leads model behavior at OpenAI, emphasized that the company is taking the feedback seriously and aims to clarify any misunderstandings. OpenAI has maintained that Sky's voice was not derived from Johansson's and has apologized for any confusion. The voice was removed from the default options after media coverage highlighted the issue.
Development and Customization of AI Voices
OpenAI CEO Sam Altman had previously approached Johansson with a proposal to license her voice for the new ChatGPT assistant. “After much consideration and for personal reasons, I declined the offer,” Johansson said through a statement to NPR. Despite this, the company proceeded with a voice that many found eerily similar. CTO Mira Murati explained that the voice was developed from another actress's voice, whose identity remains undisclosed for privacy reasons.
Johansson says that OpenAI's decision left her “shocked, angered and in disbelief that Mr. Altman would pursue a voice that sounded so eerily similar to mine that my closest friends and news outlets could not tell the difference.”
Advanced Features of GPT-4o
The GPT-4o model, which powers ChatGPT, offers advanced capabilities such as interpreting facial expressions, detecting emotions, and even singing on command. During a live demonstration, the voice assistant exhibited a flirtatious tone, raising questions about its engagement strategies. These features aim to enhance user interaction, but they also bring forth ethical concerns about the use of AI in mimicking human voices.
Johansson has voiced her concerns about the broader issue of deepfakes and the protection of personal likenesses. “In a time when we are all grappling with deepfakes and the protection of our own likeness, our own work, our own identities, I believe these are questions that deserve absolute clarity,” Johansson said.
This incident underscores the challenges that come with the rapid advancement of AI technologies. OpenAI plans to introduce additional voices to cater to diverse user preferences, ensuring a more inclusive and varied user experience.
In the midst of this controversy, Google has unveiled its own AI assistant, Project Astra, which takes a more straightforward and less playful approach compared to OpenAI's version. Experts suggest that the development of advanced AI voice assistants could lead to deeper connections between users and machines, potentially resulting in addictive interactions.
