Codex Comes to ChatGPT Mobile for On-the-Go Coding Oversight
OpenAI is extending its Codex AI coding agent into the ChatGPT mobile app, giving developers a new way to handle ChatGPT mobile coding tasks from iOS and Android devices. Rather than running code directly on the phone, the app acts as a control surface for Codex sessions already active on a laptop, Mac Mini, devbox, or managed remote environment. Users can inspect live threads, review AI-generated outputs, approve commands, switch models, and add fresh context while away from their desks. Terminal output, screenshots, diffs, tests, and approvals stream in real time, turning short decision points into quick mobile interactions instead of blocking long-running workflows. The preview is available across all Codex plans, including Free and Go, positioning the ChatGPT app as a central hub for Codex AI agent control without requiring developers to keep their primary machines constantly in front of them.
From Monitoring to Approval: Managing AI Coding Agents Remotely
The new mobile capabilities are designed for remote code management of long-duration AI workflows that don’t neatly fit into a single sitting. Many developers previously left laptops open just to supervise Codex, review diffs, or approve critical actions. Now those same checkpoints—clarifying instructions, choosing between implementation paths, or greenlighting a risky command—can be handled directly from a phone. Through the ChatGPT app, developers can converse with Codex, review generated code, and decide whether to proceed, keeping humans firmly in the loop while the AI handles repetitive or multi-step tasks such as debugging, repository navigation, and workflow automation. This model aligns with a broader industry trend: agentic AI systems that can execute complex development tasks while reserving humans for supervision and governance. By shifting these oversight tasks to mobile, OpenAI is making it easier to maintain continuous control without being physically tethered to a single workstation.
Syncing with Desktop and Remote Environments via Secure Relay and SSH
Codex on ChatGPT mobile stays in lockstep with desktop environments by loading the live state of the machine where Codex runs. Files, credentials, permissions, and local setup remain on that machine, while the phone becomes a secure remote viewport. OpenAI routes this connection through a relay layer that keeps trusted machines reachable across devices without exposing them to the public internet, reinforcing security for remote code management. On the enterprise side, Codex now supports remote SSH access, allowing it to connect into approved development environments that hold company-specific dependencies, policies, and compute resources. The Codex desktop app can detect SSH hosts from a user’s existing configuration and run projects inside those remote machines. These environments, once connected, become part of the same relay network that powers Codex AI agent control from mobile, enabling developers to jump between local and remote setups seamlessly while monitoring them via iOS Android development workflows.
Compliance, Hooks, and the Road Ahead for Mobile Codex Agents
Beyond mobile access, OpenAI is expanding Codex with features aimed at regulated and enterprise contexts. HIPAA-compliant use is supported for eligible ChatGPT Enterprise workspaces when Codex operates in local environments, making the tool viable for sensitive workflows that must meet strict data-handling standards. Hooks are now generally available, letting teams customize behavior by repository or directory, for tasks such as scanning prompts for secrets, running validators, logging conversations, or creating memories that shape how Codex responds. Programmatic access tokens bring scoped credentials into ChatGPT workspaces for Business and Enterprise users, reinforcing governance. Together, these updates signal a shift toward long-running, supervised agents that fit into existing development pipelines. With Codex now accessible through ChatGPT mobile, developers gain a unified surface to oversee AI-driven coding, approve crucial steps, and keep complex projects moving—even when they’re away from their primary machines.