OpenAI adds Chrome extension support to ChatGPT desktop browser

OpenAI just turned every Chromium-based browser into a ChatGPT co-pilot. The company’s ChatGPT desktop app now supports Chrome extensions and the Chrome DevTools Protocol, giving users the ability to interact with live browser sessions directly through the AI assistant.

The move effectively replaces OpenAI’s standalone Atlas AI browser, which was quietly retired as leadership redirected resources toward building browsing capabilities directly into the desktop app. Instead of maintaining a separate browser product, OpenAI opted to meet users where they already are: inside Chrome, Edge, Brave, and Vivaldi.

How the extension works

The ChatGPT Chrome extension, now available through the Chrome Web Store, acts as a bridge between the desktop app and whatever browser you’re already using. It enables page summarization, the ability to kick off tasks from open tabs, and context-aware conversations based on what you’re currently looking at online.

The extension works across most Chromium-based browsers, though Opera users miss out on the side chat feature.

There’s a permissions layer built in. The extension requires users to explicitly approve access for sensitive actions involving site data or page content. Page context gets shared with OpenAI when you use the feature.

The desktop app needs to be installed for the extension to function properly. Millions of users are already running the extension since it became available in early July 2026.

Developer mode and CDP access

The more interesting addition for technical users is Developer mode. This feature provides access to the Chrome DevTools Protocol, the same inspection and automation toolset that developers use to debug web applications, run performance audits, and automate browser interactions.

CDP access lets ChatGPT inspect page elements, monitor network requests, and interact with web apps at a level that goes well beyond simple page summarization. CDP access requires explicit approval on a per-site basis, and in enterprise or workspace environments, administrators control whether the feature is available at all. Organizations can restrict full CDP access through admin settings.

The rollout of these features varies depending on user plan, region, and device specifications.

Why Atlas AI got the axe

OpenAI’s Atlas AI browser had a relatively brief existence. The standalone browser was OpenAI’s initial attempt at creating a dedicated browsing experience infused with AI capabilities. Leadership ultimately decided that building and maintaining an entire browser was a misallocation of resources when the same functionality could be delivered through an extension model.

The competitive landscape

This release puts OpenAI squarely in the crosshairs of Google’s own AI browser integration efforts. Google has been steadily embedding Gemini into Chrome through its side panel feature, which offers similar contextual AI assistance while browsing.

OpenAI’s extension works across multiple Chromium browsers, not just Chrome. Users on Edge, Brave, or Vivaldi get access to the same ChatGPT browsing features. The CDP access for developers is also a differentiator that Google hasn’t matched with Gemini’s browser integration.

For enterprise customers, the admin controls and workspace-level policy management allow organizations to govern which employees can share page context or access CDP features.

Microsoft’s Copilot in Edge represents another competitor in this space, though OpenAI’s partnership with Microsoft adds complexity to that dynamic.