OpenAI's ChatGPT Chrome extension can now read your open tabs and control them via ChatGPT Work

OpenAI has upgraded its ChatGPT extension for Google Chrome so ChatGPT can read the tabs you already have open, according to The Verge. With the extension installed, users can tag Chrome in ChatGPT's web or desktop app and have the assistant pull information from those tabs. The report, published October 15, 2026, says the ChatGPT Atlas browser is no longer available, and the extension is now the route for browser-based features.

What's new

Tab access. When a user tags Chrome (@chrome) in the ChatGPT web or desktop app, ChatGPT can view the user's open tabs. OpenAI's demo video shows a user asking ChatGPT to compare sofas across three open tabs.

Sidebar. The extension now brings ChatGPT into a Chrome sidebar. OpenAI says this includes "the full power of ChatGPT" and the user's entire conversation history in Chat or Work, so users can keep working without leaving the page they are viewing.

Tab control via ChatGPT Work. OpenAI says ChatGPT Work can now "control" Chrome tabs for tasks such as sending messages, filling out forms, or downloading files. It can also organize and group open tabs.

Example prompts from OpenAI

OpenAI listed several sample queries for the upgraded extension:

• Use the details in an event brief to fill out a registration form, stopping before submission.
• Put vendors from open tabs into a comparison sheet with pricing, key features, and source links.
• Compare contractor services across open tabs, then draft an email to the best fit for a home renovation.
• Fill out an open form ahead of a dentist appointment and add the appointment to work and personal calendars.

The prompts show the intended pattern: ChatGPT reads multiple tabs, then acts on them with a handoff to calendars, sheets, or email. Two of the examples include a stop-before-submit instruction, which suggests users can limit how far the agent goes.

What's not disclosed

The source does not say which model powers these features. It also gives no pricing, plan or tier availability, regional rollout, or rate limits. Whether tab control is on by default or requires per-action approval is also not specified. No benchmark results or reliability figures for the browser-control tasks have been published.

The Verge compares ChatGPT Work's tab control to Google's Gemini-powered "auto browse" feature in Chrome. That comparison is the publication's framing, not an OpenAI claim.

What this means

OpenAI has moved from shipping its own browser to embedding in the one most users already run. Atlas required users to switch browsers, so the Chrome extension removes the biggest adoption barrier. The trade-off is that OpenAI now works inside Google's platform, within whatever the Chrome extension model allows.

The larger change is the shift from reading a single page to working across tabs. Tab-level context makes tasks like comparison shopping and vendor research much easier, and it is the kind of work agents handle best. Form-filling and file downloads carry higher risk. Misread fields, prompt injection from untrusted page content, and unintended submissions are the failure modes to watch, and OpenAI has not yet detailed its safeguards.

For teams evaluating browser agents, the missing details matter most: plan eligibility, permission prompts, and how much page content is sent to OpenAI's servers. Until those are documented, treat tab control as a feature to pilot on low-stakes workflows.