ChatGPT's Chrome Extension Now Reads Your Open Tabs

OpenAI has added a capability to ChatGPT's Google Chrome extension that lets the assistant read the browser tabs a user already has open, the company said, showing the feature in a demo video. The change means someone working in ChatGPT's web or desktop app can tag Chrome and have the model pull in whatever is on screen in their browser, without copying text or pasting URLs by hand. It is the second time this year that OpenAI has tried to put ChatGPT directly into the browsing session, after its Atlas browser effort was shut down.

The mechanics are straightforward. With the extension installed, a user references Chrome inside ChatGPT, and the assistant gains visibility into the tabs that are currently open. OpenAI's demo shows the model working across that material rather than waiting for a single pasted excerpt. The extension had already existed before this update; what changed is how much of the browsing context the assistant can reach on its own.

That matters because context has become the main constraint on what chat assistants can actually do. A model that can only see what you paste is limited to one document at a time. A model that can see every tab you have open is closer to a working assistant that follows a task across several sources. OpenAI's move follows the same logic that drove browser makers and rival labs to build agentic browsing features, where the assistant acts on the page instead of just talking about it.

The timing is notable given Atlas. OpenAI had pitched a dedicated browser as a way to give ChatGPT a native home on the web, and that product is now gone. Folding the same tab-awareness into an extension for Chrome, the browser most people already use, is a quieter path to the same goal. It also avoids asking users to switch browsers before they can try the feature.

Tab access is a sensitive permission, and OpenAI has not said in detail how long tab data is retained, whether it is used for training, or how users revoke access once the extension has it. Those questions will shape whether people leave the permission on. The demo video shows the feature working; the privacy documentation around it will decide how many users keep it enabled.

Watch for how OpenAI scopes this next. Reading open tabs is a short step from acting on them, and the company has been moving toward agents that complete tasks rather than just summarize them. Whether the Chrome extension becomes a full browsing agent, or stays a context pipe, is the open question.