OpenAI's ChatGPT Chrome extension can now read the tabs you have open and use them as context in a chat — compare pages and draft from what you are reading — and, in Work mode, act in the browser.
Last verified · 2026-10-17 · by Moe Ameen
ChatGPT Chrome tab access is a capability of OpenAI's ChatGPT browser extension, broadened in mid-October 2026. With the extension installed, ChatGPT can look across the tabs you have open and use them as context inside a chat: you ask a question and it answers from the pages that are loaded, so you can compare several listings, pricing pages, or articles without pasting their text in. OpenAI described the wider update as bringing the power of "computer use" to anyone who uses Chrome.
It joins features the extension already shipped: a side-chat panel that sits next to your browsing, the ability to ask about a YouTube video you have open, and a highlight-text / right-click "Ask ChatGPT" shortcut. The extension is free in the Chrome Web Store, and OpenAI says it is available on all ChatGPT plans. On the desktop app it runs beyond Chrome too — Edge, Brave, Opera, and Vivaldi — though side chat is not available in Opera.
The more aggressive half is agentic. In Work mode — OpenAI's agent for longer tasks — ChatGPT can control Chrome: send messages, fill out forms, download files, and group and organize your open tabs. It asks before interacting with each new site (Allow once, for this site, for all sites, or Decline), and flags "Allow for all sites" and browser-history access as elevated risk, with history granted per request only.
Read as a creator, the feature is your intake and research layer. It reads the web you are looking at, compares it, drafts from it, and can operate the browser — but it renders no branded video, images, or carousels, holds no brand voice you configure once, sizes nothing per platform, and publishes nothing to your social feeds. That seam is exactly where a content engine picks up.
The honest way to use ChatGPT's tab access is as the read half of a two-stage line, and to let [Kompozy](/) be the render-and-ship half. ChatGPT reads what is on your screen; Kompozy renders what your audience sees. Stage one lives in the browser: you keep a competitor's YouTube tab, a product page, and two trend articles open, and ChatGPT compares them and hands you a script, an outline, or a caption angle drawn straight from those pages. That is a sharper intake step than copy-pasting into a blank chat. Stage two is everything the browser can't do — because a chat answer and an in-browser action are not posts.
Run it as a handoff. Take the script or outline ChatGPT synthesized from your open tabs and drop it into Kompozy's Quick Ingest as the seed for a content unit. Kompozy fans that one seed into the formats a browser assistant never renders: a captioned [Persona Shorts](/glossary/persona-shorts) or HeyGen avatar video with one consistent face, a [Persona Frames](/glossary/persona-frames) clip composited into a brand-exact template, a [Clipped Short](/glossary/clipped-short) cut from long-form video, [Carousel](/glossary/hyperframes) posts and quote graphics, photo posts, per-platform text posts, a blog, and a newsletter — every piece held to your voice by a [Persona Brief](/glossary/persona-brief) and banned-word filters, and rendered as finished media rather than a description of it. Then [Autopilot](/glossary/autopilot) schedules and publishes the set across the eight social platforms plus blog and email behind a per-post review. Keep the reading-and-comparing loop you like in Chrome; let Kompozy be the part that actually produces and posts. For related surfaces see [ChatGPT](/ai-tools/chatgpt) and [ChatGPT Apps & Automations](/ai-tools/chatgpt-apps-and-automations).
It is a capability of OpenAI's ChatGPT Chrome extension that lets ChatGPT read the tabs you have open and use them as context in a chat — so you can ask it to compare several pages or draft from what you are reading without pasting the text in. OpenAI broadened it in mid-October 2026 alongside Work-mode controls that let ChatGPT act in the browser.
No. Reading your tabs produces a chat answer, and Work mode performs browser actions like filling a form or organizing tabs. Neither renders finished on-brand video, carousels, or graphics, holds a persistent brand voice, or publishes to your social feeds. To turn what it found into finished posts and publish them across platforms, you use a content engine like Kompozy.
Treat it as the research front. Keep your sources open — a competitor video, a product page, a few articles — have ChatGPT compare them and draft an angle or script, then hand that to Kompozy, which renders the video, carousels, images, blog, and newsletter and schedules them across platforms. ChatGPT reads and drafts; Kompozy produces and ships.
OpenAI says the updated extension is available on all ChatGPT plans and is free in the Chrome Web Store. Beyond Chrome, the desktop app lists Edge, Brave, Opera, and Vivaldi, with side chat not available in Opera. Rollout and some agentic features can vary by account and region, so confirm current details on OpenAI's pages.
They are two halves of the same update. Reading tabs uses your open pages as context for a chat; the Work-mode controls let ChatGPT act in Chrome — send messages, fill forms, download files, organize tabs — with per-site permission prompts and elevated-risk flags for broad grants. One is context in, the other is action in the browser; neither posts content to your feeds.