OpenAI Launches Voice Mode for ChatGPT

2024-07-31

OpenAI has started rolling out a new advanced voice mode for ChatGPT to a select group of ChatGPT Plus subscribers. This feature was first unveiled during the GPT-4o launch event held by OpenAI in May, but it faced criticism for sounding similar to Scarlett Johansson's voice and was delayed for safety reasons. During OpenAI's event, the new voice mode was noticeably more powerful than ChatGPT's current voice mode. OpenAI employees were able to interrupt and request the chatbot to tell stories in different ways, and the chatbot could handle these interruptions and adjust its responses accordingly. The advanced mode was originally scheduled to be released in alpha version by the end of June, but OpenAI postponed it for a month to "meet our release standards." As part of the delay, the company stated that it is improving the model's ability to detect and reject certain content. OpenAI spokesperson Taya Christianson mentioned that the company has tested the voice model's capabilities with over 100 external red team members (individuals who attempt to find vulnerabilities by attacking the technology). Given the recent scrutiny of OpenAI's security policies, this pause may be the right choice. Christianson also stated that OpenAI has "added new filters that can identify and block requests for generating copyrighted music or other protected audio." One of the main criticisms of the new mode during OpenAI's event was that the voice demonstrated, called "Sky," sounded very similar to Scarlett Johansson's AI character in the movie "Her." Although this voice had already existed in ChatGPT prior to the spring demo, OpenAI withdrew it shortly after Johansson revealed that she had reached out to inquire about how the voice was created. Christianson stated that ChatGPT's new mode will only use four preset voices created in collaboration with voice actors and added, "We have ensured that ChatGPT cannot mimic the voices of others, including individuals and public figures, and will prevent output that differs from these preset voices." According to Christianson, OpenAI plans to roll out this new mode to all ChatGPT Plus users in the fall of this year.