Products & Tools

ChatGPT's Chrome Extension Can Now Read Your Tabs and History

OpenAI's ChatGPT Chrome extension can now read open tabs, access full ChatGPT history in the sidebar, and let the Work agent send messages and fill out forms — free on all plans.

ChatGPT can now look at your Chrome tabs to do tasks for you
ChatGPT can now look at your Chrome tabs to do tasks for youVMonte13 / pixabay
By James Calloway5 min read

Updated

Why it matters

  • OpenAI announced the tab-reading ChatGPT Chrome extension update on October 15, 2026.
  • OpenAI retired its own browser on August 9, 2026.
  • The extension's Work mode can send messages, fill out forms and organize tabs.
  • The sidebar now gives access to a user's entire ChatGPT history.
  • The extension is free on all ChatGPT plans and works in Edge, Brave and Opera.

OpenAI's ChatGPT extension for Chrome can now read your open tabs, and in its Work mode it can send messages, fill out forms and organize those tabs on your behalf. The update, announced by OpenAI's official ChatGPT account on X on October 15, 2026, arrives roughly two months after the company retired its own browser on August 9. OpenAI is now competing for browser real estate from inside rival software rather than from its own product.

The move matters because the browser is the primary interface between users and the web — and, increasingly, between users and AI assistants. Google ships Gemini directly into Chrome, the world's dominant browser. OpenAI, having folded its own browser effort, is now betting it can reach the same users through an extension that works not only in Chrome but also in other Chromium-based browsers, including Microsoft Edge, Brave and Opera.

What can the extension actually do?

The core upgrade is tab access. In ChatGPT Chat, users can ask the assistant to look at open Chrome tabs and generate answers grounded in what is on screen. OpenAI's own example is comparison shopping: with three couch listings open in three tabs, a user can ask ChatGPT to compare the products by dimensions, prices, delivery time and return policy.

That is a concrete, verifiable use case, and it is the strongest of the examples the company offers. The others lean on ChatGPT Work, OpenAI's agent mode designed for complex multi-step tasks. There, the company says, the assistant can:

  • send messages;
  • fill out forms;
  • organize tabs.

OpenAI repeats "fill out forms" twice in its announcement, which suggests either a copy-editing slip or that form-filling is the capability it most wants users to notice.

The Work examples are more ambitious. The company suggests asking ChatGPT to fill out a form for a new dentist and then add the resulting appointment to the user's calendar. A second example has the assistant analyze several contractors' terms spread across multiple tabs and draft an email to the provider that best fits the user's needs.

These are agentic tasks — actions taken on the user's behalf across steps, not just answers to questions. They are also the tasks most likely to fail in ways users notice, because they involve the AI both reading correctly and acting correctly. OpenAI's announcement does not say what happens when Work fills out a form incorrectly or sends a message a user did not intend. Takeaway: the promised capabilities are useful in principle, but the announcement is silent on error handling and undo.

What about privacy and history access?

The second significant change is access to browsing history. OpenAI says the ChatGPT sidebar in Chrome now has access to the user's entire ChatGPT history, so tasks can be assigned without switching away from the current page. A user can keep a site open on one side and run Chat or Work in the sidebar alongside it.

Note the precise wording: the announcement describes access to ChatGPT history, not Chrome browsing history. The company has not stated, in this announcement, how the new tab-reading feature interacts with page content, how long tab contents are retained, or whether any of it is used for training. Those are the questions users typically ask first when an AI assistant gains eyes on open pages, and OpenAI has not answered them here.

What OpenAI does say is availability: the extension works regardless of which ChatGPT plan a user has. There is no paid tier gating the new features, which distinguishes the rollout from other OpenAI product launches that have reserved advanced capabilities for subscribers. Takeaway: the extension is free to all plans, but the announcement leaves data-handling details unspecified.

Why did OpenAI retire its own browser?

The August 9 retirement of OpenAI's browser is the necessary context for this update. Building a browser means competing head-on with Google, whose Chrome holds the majority of the desktop browser market and which can integrate its own assistant at a level no extension can match. An extension strategy inverts that fight: instead of asking users to switch browsers, OpenAI is asking them to install one piece of software inside the browser they already use.\n The company frames the earlier extension upgrades as having made ChatGPT "a more viable alternative to Gemini in the browser." That is OpenAI's own competitive assessment, and it should be read as such. The extension now covers Chat for reading and answering, Work for acting, and sidebar access for continuity — three surfaces where Gemini and other assistants also operate.

Whether parity in features translates into parity in adoption is a separate question. Gemini benefits from being preinstalled in Chrome; ChatGPT requires an explicit install. OpenAI has not released usage figures for the extension, so there is no way to verify the "viable alternative" claim from the outside. Takeaway: the extension strategy is a pragmatic retreat from the browser war, but its success depends on numbers OpenAI has not published.

What should users watch next?

The October 15 announcement positions ChatGPT as an operator inside the browser rather than a chatbot pointed at it. The gap between those two roles is where the risk sits. An assistant that compares couches across tabs costs a user nothing if it reads a price wrong; an assistant that sends messages and fills out forms does not enjoy the same margin for error.

Three details will determine whether this update matters beyond launch week:

  • whether OpenAI publishes data-retention and training policies for tab and page content;
  • whether Work-mode actions include review steps before messages are sent or forms are submitted;
  • whether the extension's cross-browser support in Edge, Brave and Opera delivers feature parity with Chrome.

None of the three is addressed in the announcement. Until they are, the update is best understood as a capability claim with its safety and privacy terms still unwritten — a pattern that has preceded most of OpenAI's agent-mode releases, and one this announcement does not break.

Original: chromewebstore.google.com

Share this article:

More from James Calloway

James Calloway

Show full bio

News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.

285 articles

Related articles

  1. ChatGPT's Chrome extension can now read your open browser tabs
  2. OpenAI launches ChatGPT Atlas, a browser with AI at its core
  3. OpenAI Built a New Browser Architecture Called OWL for ChatGPT Atlas
  4. OpenAI Turns ChatGPT Into an App Store Challenger
  5. OpenAI Turns ChatGPT Into an App Platform With Plugin Overhaul

« Previous article