OpenAI Codex Now Runs Apps Autonomously in Background

This isn’t just an update. It’s a declaration. OpenAI has taken Codex from a passive assistant to an active, persistent agent that lives inside your operating system.

It runs in the background, schedules its own work, and wakes itself up. It edits images, browses the web, and plugs into GitLab, Atlassian, and Microsoft Suite. The message to Anthropic is clear: Claude Code may have set the pace, but OpenAI is now playing a different game entirely.

The real shift here is autonomy. Codex no longer waits for your command. It works alongside you, parallel to you, even while you sleep.

For developers, that means testing, iterating, and automating without context-switching. For the rest of us, it signals a future where AI agents don’t just answer questions, they execute tasks, manage workflows, and persist across sessions. The background is no longer a passive space.

It’s a workspace. But the rollout is cautious. macOS first.

EU users wait. That restraint matters. It suggests OpenAI knows the stakes.

Giving an agent the keys to your desktop, the ability to wake itself up, and the power to run multiple instances in parallel is a leap. One bug, one misstep, and the background becomes a liability. Still, the direction is undeniable.

The era of the single, reactive chatbot is over. The new frontier is persistent, parallel, and proactive. Codex is no longer a tool you use.

It’s a colleague you deploy. And with that, the race between OpenAI and Anthropic has shifted from who can answer faster to who can act more reliably. The winner won’t be the one with the best conversation.

It will be the one you trust to work while you walk away.