OpenAI Updates ChatGPT Desktop App with Voice Commands to Control Computer Tasks

On August 9, OpenAI announced an update to the ChatGPT desktop app, adding support for ChatGPT Voice, which enables users to converse with ChatGPT via voice and control AI agents to perform tasks on their computers. This feature is powered by OpenAI’s newly launched ChatGPT-Live series of voice models and supports both ChatGPT Work and Codex, with the ability to invoke computer operation capabilities to access websites and applications. On macOS, users can grant ChatGPT permission to access screen content, including alt-text. The desktop version allows users to issue complex, multi-step voice commands and respond promptly to ChatGPT’s requests or confirmations. OpenAI demonstrated an example in which a developer used a single voice command to instruct ChatGPT to create a new code thread, submit a Pull Request, and identify the root cause of a bug. Additionally, users can access ChatGPT Voice in Codex via the iOS app’s remote access feature. Meanwhile, Anthropic has updated Claude’s voice mode, which leverages the Opus, Sonnet, and Haiku models to carry out tasks within apps such as Gmail, Calendar, Slack, Notion, and Canva.