ChatGPT Voice now controls desktop Codex and multi-agent workflows
Original: ChatGPT Voice reaches desktop control for Codex and multi-agent work View original →
Voice becomes a work-control layer
OpenAI has moved ChatGPT Voice into the desktop app, extending it from spoken conversation into computer control and agent coordination. In the tweet, OpenAI said users can
“direct multiple agents running in ChatGPT Work or Codex”with voice, with global rollout on macOS and Windows for Plus, Pro, Business, Edu, and Enterprise plans. The post passed 2.6 million views and 10,000 likes soon after publication.
The technical hinge is GPT-Live. OpenAI says the system can speak, listen, and coordinate work inside the app at the same time. That matters because desktop agent work is often constrained by context switching: reading code, issuing instructions, watching execution, then redirecting when a tool stalls. A voice layer could compress those loops if permission checks and task state remain visible.
The OpenAI account is typically used for product availability, model rollouts, and enterprise workflow updates, so this post is more than a casual demo. It places Codex and ChatGPT Work inside the same interaction pattern: users speak, the app coordinates, and multiple agents can continue running. A follow-up post added that ChatGPT Voice can also reach Codex from the iOS app using paired remote access, with Android support still pending.
What to watch next is governance. Desktop control through voice raises harder questions than voice chat: which actions require confirmation, how enterprise admins audit instructions, and whether the system can recover cleanly when an agent misreads intent. The source tweet is available here.
Related Articles
Sam Altman said usage of OpenAI's agentic products rose 2.5x in a week, a sharp adoption signal for Codex and ChatGPT Work. The number matters because agent workflows are moving from demos into recurring work.
OpenAI’s new Economic Research paper frames Codex as a shift from chat to delegated work. Among organizational users, non-developer Codex adoption is up 189x since August 2025, and 70.2% of sampled individual users have assigned at least one task estimated to take a human more than an hour.
Agentic tools are moving from coding demos into internal operating workflows. OpenAI says people across the company use Codex for more complex, longer-running, cross-functional work, and the post drew more than 1.1 million views on FxTwitter.