OpenAI Postpones Launch of New ChatGPT Voice Mode by Over One Month

OpenAI's Spring Update event in May 2024 was highlighted by the introduction of GPT-4o and the highly anticipated Voice Mode for ChatGPT. This new feature aims to create a more natural, conversational interaction with the AI, complete with emotional inflection and the ability to manage interruptions. However, the rollout of Voice Mode has been delayed by at least a month, pushing expectations to late July or early August at the earliest. Initially, it will only be accessible to a small group of users.

OpenAI announced this delay via its X account, indicating the need for additional time to refine the Voice Mode. The company emphasized improvements in the model’s ability to detect and refuse inappropriate content.

The statement read: “We had planned to roll out Voice Mode in alpha for a select group of ChatGPT Plus users in late June, but we need one more month to meet our quality standards. We're enhancing content detection capabilities, improving user experience, and ensuring our infrastructure can scale effectively while delivering real-time responses. Our phased deployment strategy involves gathering feedback from an initial user group before expanding access to all Plus users in the fall, contingent on meeting our safety and reliability benchmarks.”

Voice Mode promises to enable AI to understand and respond with emotional and non-verbal cues, enhancing the conversational experience. OpenAI is committed to thoughtfully introducing these advancements.

This delay poses challenges for OpenAI as it competes with rivals like Anthropic, which recently launched the Claude 3.5 Sonnet foundation model, reportedly outpacing GPT-4o in several third-party assessments. OpenAI also faces scrutiny over its safety measures in pursuit of achieving artificial general intelligence (AGI), leading to internal criticism regarding restrictive separation agreements and equity limits—many of which have now been amended.

External criticism continues, notably from actress Scarlett Johansson, who protested against the AI voice “Sky” that she felt resembled her own without her consent. OpenAI responded by clarifying that the voice actress for Sky was approached separately from Johansson, though the company has since disabled that AI voice.

Despite these challenges, OpenAI continues to attract new users and partnerships, including collaborations in the creative industry with its unreleased video AI model Sora, and integrating GPT-4o into healthcare applications like a cancer screening tool for Color, alongside securing numerous enterprise clients.