After rival Anthropic's Claude Code debuted mobile control, OpenAI has now added the same capability to ChatGPT. You can now remotely monitor and command Codex tasks running on your desktop or cloud environment. The company also revealed that Codex has surpassed 4 million weekly active users.
The mobile interface doesn't just send simple commands — it syncs the real-time state of your development machine. Whether Codex is running locally or on a remote server, your phone can display terminal output, screenshots, test results, and code diffs. As AI agents take on longer tasks, you can use spare moments to make decisions or approve permissions, preventing delays while waiting for manual input.
The cross-device connection is powered by a secure relay layer, so your development machine doesn't need to be exposed to the public internet. For enterprise environments, Remote SSH allows direct connections to internal servers, and Hooks let you customize agent behavior and filter sensitive information. Enterprise plans also add HIPAA compliance for running Codex locally.
As large language models become capable of executing long, autonomous tasks, the bottleneck in human-machine collaboration has shifted from writing code to reviewing and correcting it. Whether with Claude Code or Codex, putting the console on your phone and capturing developers' spare attention is becoming the standard workflow for AI agent tools.