ChatGPT Desktop App Gets Voice Control for AI Agents

OpenAI has updated its ChatGPT desktop app with new voice capabilities. This allows users to control AI agents and perform complex computer tasks using spoken commands. The feature leverages the company’s new GPT-Live voice model family. ChatGPT Voice now works across both ChatGPT Work and Codex environments. Users can dictate multi-step commands that the AI can execute autonomously. For example, a developer can use a single voice command to create a new thread, make a pull request, and identify a bug. The system can also navigate websites and applications. On macOS, the feature integrates with Appshots, giving the app access to on-screen content for more context-aware assistance. This desktop update significantly expands on the smartphone version, which focused more on natural conversation rather than direct actions on the device. Competitors like Anthropic are also updating their voice modes for AI assistants like Claude. Both companies aim to make voice a primary interface for AI-assisted work. This shift to hands-free, conversational computing could reduce friction in complex workflows for developers and knowledge workers. It also signals that voice is becoming a core interaction method for professional users.