Odaily News OpenAI announced that it has begun rolling out advanced voice mode to a small number of ChatGPT Plus users. The mode provides more natural real-time conversations, allows interruptions at any time, and is able to sense and respond to users' emotions. Users participating in this test will receive instructions via email and mobile app. OpenAI plans to gradually open the advanced voice mode to all Plus users by the fall. Video and screen sharing features will be launched later. OpenAI said that after the first demonstration of the advanced voice mode, it has been strengthening the security and quality of voice conversations in order to bring this cutting-edge technology to millions of people. OpenAI tested the voice function of GPT-4o with more than 100 external testers, covering 45 languages. To protect privacy, the model will only use four preset voices, and establish a system to block other voice outputs, while implementing protection measures to prevent violent or copyrighted content. OpenAI said it will learn from this test to make the advanced voice experience safer and more enjoyable, and plans to share a detailed report in early August to introduce the capabilities, limitations and security assessments of GPT-4o.