OpenAI has upgraded its ChatGPT browser extension, allowing the AI assistant to use information from users’ open browser tabs and perform tasks directly through a supported browser setup. Announced on October 15, 2026, the update brings browser awareness and computer-use capabilities to people who want to work with existing websites rather than repeatedly copy information into a chat. Users can ask ChatGPT to compare products across tabs, summarise articles, examine subscription plans or gather information from several pages.
The update goes beyond answering questions about websites. In ChatGPT Work mode, the assistant can interact with browser tabs to fill out forms, send messages, download files and organise tabs into groups. The capabilities depend on the browser integration, permissions and account configuration. OpenAI’s move also signals a broader strategy: making ChatGPT an assistant that can work within the tools people already use, rather than limiting it to conversations in a standalone chatbot window.
Key Takeaways
- Open tabs as context: ChatGPT can use information from multiple open browser tabs to answer questions.
- Two modes: Chat mode focuses on understanding and discussing pages, while Work mode can carry out browser tasks.
- Task automation: Supported tasks include filling forms, sending messages, downloading files and organising tabs.
- Browser support: The official integration supports Chrome, Edge, Brave, Opera and Vivaldi, with some differences in features.
- Desktop setup: Users need the appropriate ChatGPT desktop app configuration and official browser extension for browser control.
- Privacy considerations: Browser access can expose information from signed-in websites, so users should review permissions and supervise important actions.
- Gradual rollout: Availability can vary by account, browser, workspace settings and rollout status.
What Has OpenAI Changed?
Previously, people often had to copy text, paste links or describe a webpage manually before ChatGPT could analyse it. The updated browser integration is designed to reduce that friction by allowing ChatGPT to work with information already available in the user’s browser.
For example, a shopper comparing three laptops can ask ChatGPT to examine the open product pages and compare specifications, prices, warranties and delivery terms. Someone researching a business story can ask it to compare several reports without manually transferring each article into the conversation.
The assistant can use information across tabs to create a more complete answer. This can be useful when different websites contain different pieces of the same puzzle, such as product specifications on one page, shipping policies on another and customer support information on a third.
OpenAI has also expanded ChatGPT’s role within the browser itself. The updated extension provides access to ChatGPT conversations and Work capabilities alongside the website being viewed, helping users continue their work without constantly switching between applications.
The result is a closer connection between conversational AI and everyday browsing. Rather than only explaining what a user should do, ChatGPT can help carry out selected steps on supported websites.
Chat Mode vs Work Mode: What Is the Difference?
The new capabilities are divided into two broad workflows: understanding browser content and acting on that content.
| Capability | Chat mode | Work mode |
|---|---|---|
| Read information from open tabs | Yes, when the browser is selected and access is available | Yes |
| Compare information across websites | Yes | Yes |
| Explain an article or webpage | Yes | Yes |
| Fill out online forms | Not its primary role | Supported |
| Send messages through a website | Not its primary role | Supported, subject to permissions |
| Download files | Not its primary role | Supported |
| Organise or group tabs | Not its primary role | Supported |
Availability and specific actions depend on the supported browser integration, account settings and permissions.
Chat mode uses your tabs to answer questions
Chat mode is designed for research, comparison and explanation. Users can select or mention their browser or open tab and ask questions about the content.
A person researching a smartphone purchase, for example, could open several product pages and ask ChatGPT to compare battery capacity, camera specifications, prices and return policies. The assistant can use the available page information to organise the comparison and explain the differences.
The same approach can help with business research. A user following a company’s quarterly results could ask ChatGPT to compare revenue figures, management commentary and analyst reports across several open pages.
The main benefit is convenience: users spend less time switching windows and copying text. However, ChatGPT can only make a reliable comparison when the relevant information is accessible and correctly interpreted. Missing details, inconsistent figures or outdated pages can still affect the result.
Work mode can interact with websites
Work mode takes the process further by allowing ChatGPT to operate supported browser tabs to complete tasks.
Instead of merely explaining how to complete an online registration form, the assistant can help fill in fields. It can also perform supported actions such as sending messages, downloading files and organising open tabs.
This capability can be useful for repetitive administrative work. A person who regularly collects information from several websites could ask ChatGPT to gather the details, organise them and prepare a structured result.
However, browser control is not the same as unlimited access to every website or permission to perform every action automatically. Access is governed by the configured browser integration, website permissions and safety checks. Users should review the information being entered and the destination before allowing consequential actions to proceed.
Which Browsers Does ChatGPT Support?
OpenAI’s browser integration supports several major browsers through its official extension and desktop workflow.
| Browser | Browser tab integration | Side chat |
|---|---|---|
| Google Chrome | Supported | Supported |
| Microsoft Edge | Supported | Supported |
| Brave | Supported | Supported |
| Opera | Supported | Not supported in the same side-chat workflow |
| Vivaldi | Supported | Supported |
Support reflects OpenAI’s browser integration documentation. Exact availability can depend on the desktop app version, browser extension and account configuration.
This broader compatibility matters because users do not all browse the web in the same way. Some prefer Chrome for its extensions, while others use Edge, Brave or Vivaldi for their own workflows.
OpenAI’s approach lets the assistant use an existing browser profile where supported, which can be useful for websites where a person is already signed in. That convenience also makes permission management particularly important, since signed-in pages may contain personal or confidential information.
Users should install the official extension and follow the setup instructions in OpenAI’s documentation rather than granting access to unverified third-party extensions.
How to Set Up ChatGPT With Your Browser
The setup process depends on whether a user wants to discuss pages or let ChatGPT interact with them.
To use the supported browser integration:
- Update the ChatGPT desktop app. Make sure the application is current.
- Open the app’s Work settings. Locate the browser or computer-use integration options.
- Install the official browser extension. Follow OpenAI’s instructions for the browser you use.
- Approve the requested permissions. Review what the extension can access before enabling it.
- Choose your browser in ChatGPT. Start a Work or Codex task and select the relevant browser through the available browser mention or selection menu.
- Give a specific instruction. For example, ask ChatGPT to compare three open product pages or collect specified details from several websites.
- Review the result. Check the extracted information and supervise any actions that change a website or send information elsewhere.
Users who only want help with the current webpage can also use the browser’s ChatGPT side-chat option where supported.
If the feature is not available, the reason may be an incomplete setup, an older application version, a workspace restriction or a rollout that has not reached the account. OpenAI’s browser extension documentation provides the latest setup and troubleshooting details.
What Can Users Actually Do With the New Feature?
The update has practical applications across shopping, research, administration and professional work.
Compare products without copying specifications
A user shopping for furniture, electronics or household appliances can open several product pages and ask ChatGPT to compare price, size, material, warranty and delivery terms.
The assistant can turn scattered information into a single comparison. This can make decisions easier when each retailer presents details differently.
Research companies and business news
People tracking financial results or technology developments can open multiple reports and ask ChatGPT to identify what the sources agree on, where their figures differ and which claims need further verification.
This can help speed up research, but the assistant should not treat every webpage as authoritative. Company filings, official announcements and reliable reporting may deserve more weight than unsourced commentary.
Handle repetitive online forms
Work mode can help fill out supported forms using information provided by the user or gathered during the task. This may be useful for routine registrations and administrative processes.
Users should check every field carefully before submitting a form. Errors in names, dates, addresses or account details can have consequences, and a completed form should not be treated as automatically correct.
Organise a crowded browser
People often accumulate tabs while researching a topic. ChatGPT can help organise tabs into groups, making it easier to separate work projects, shopping research and general reading.
This is a relatively low-risk use case compared with sending messages or changing information on a website, although users should still review the proposed organisation before accepting it.
Privacy and Security: What Should Users Know?
The ability to read open tabs is powerful because a browser can contain much more than public information. Users may have email, financial accounts, work dashboards, private documents and other sensitive pages open at the same time.
When ChatGPT accesses a browser page, information from that page may become part of the context used to answer a question or complete a task. OpenAI’s documentation says it does not maintain a separate complete record of all browser actions through the extension, but browser content that enters ChatGPT context may be stored according to the applicable ChatGPT data controls.
Users should therefore avoid exposing sensitive pages unnecessarily. It is sensible to close unrelated confidential tabs, review permissions and make sure the correct account and website are selected before starting a task.
The risk of malicious webpage instructions
Browser agents face a security problem known as prompt injection. A malicious webpage can contain instructions designed to manipulate an AI assistant into ignoring the user’s goal or attempting an unintended action.
For example, a page might include hidden or misleading instructions asking an assistant to reveal information, visit another site or perform an action unrelated to the user’s request.
OpenAI uses permission prompts and other safeguards to reduce these risks, but no protection can eliminate every possibility of a mistake. Users should inspect unfamiliar websites, avoid exposing secrets and review the assistant’s proposed actions before approving sensitive steps.
Passwords, one-time security codes and payment credentials should never be pasted into a chat as a shortcut. Use the appropriate secure sign-in process and remain present when a task involves important account changes or transactions.
How This Changes the Competition in AI Browsers
The update places ChatGPT more directly in competition with AI tools that help users navigate and act on the web.
Google has been adding AI capabilities to Chrome and its broader ecosystem, while other AI companies are developing assistants that can interpret webpages and automate browser tasks. Users increasingly expect an AI assistant to do more than provide an answer: they want it to gather information, compare options and complete routine work.
OpenAI’s strategy is to bring these capabilities into familiar browsing environments. This can lower the barrier to adoption because people do not have to replace their preferred browser to access AI assistance.
The strategy also expands the potential role of ChatGPT from a conversational interface into a general-purpose work assistant. Browser access complements other tools that can work with documents, spreadsheets, applications and connected services.
The challenge will be reliability. Users may appreciate a feature that saves time, but they will also expect it to handle website layouts, account permissions and ambiguous instructions without making costly mistakes.
The Bigger Picture
OpenAI’s browser update represents a shift from AI that primarily answers questions to AI that can use the same online environment as the user. Reading across tabs makes research and comparisons more convenient, while Work mode adds the ability to complete supported tasks such as filling forms, sending messages and downloading files. The feature could reduce repetitive work for people who spend much of their day researching or interacting with websites.
The change also raises the importance of privacy, security and user oversight. A browser assistant may encounter confidential information and untrusted webpage instructions, so its usefulness depends on more than its ability to click or type. Clear permissions, careful review and reliable execution will determine whether browser automation becomes a dependable part of everyday work.
Looking Ahead
OpenAI’s next challenge is likely to be making browser-based assistance reliable across more websites, workflows and account configurations. The current update already supports a range of browsers, but the exact features available to users can vary. Wider adoption will depend on how easily people can set up the integration and whether it consistently completes tasks as instructed.
For users, the practical starting point is to try the feature on low-risk work, such as comparing public product pages or organising research tabs. More consequential tasks should involve closer supervision, especially when forms contain personal data or actions could send messages, change records or affect accounts. Browser control can save time, but the user remains responsible for reviewing important decisions and outcomes.
Frequently Asked Questions
1. Can ChatGPT read all my open Chrome tabs?
ChatGPT can use information from open tabs when you select the supported browser integration and grant the necessary access. The exact behaviour depends on the setup and permissions. This does not mean every tab is automatically read in every conversation.
2. Can ChatGPT fill out forms and send messages?
Yes. In Work mode, the supported browser integration can perform tasks such as filling out forms and sending messages. Users should review the content, recipient and destination before allowing consequential actions.
3. Does the feature work only in Google Chrome?
No. OpenAI’s browser integration also supports Microsoft Edge, Brave, Opera and Vivaldi, although side-chat availability differs by browser.
4. Do I need the ChatGPT desktop app?
The documented browser-control workflow uses the ChatGPT desktop app together with the official browser extension. Setup and feature availability can vary by account and workspace.
5. Is it safe to let ChatGPT control my browser?
The feature includes permissions and safety checks, but users should still exercise caution. Avoid exposing sensitive tabs unnecessarily, be careful with unfamiliar websites and review actions before submitting forms or sending information.
6. Does ChatGPT store everything I do in Chrome?
OpenAI says the extension does not maintain a separate complete record of all browser actions. However, information that enters ChatGPT context—including page content, screenshots, tool calls or summaries—may be handled under the applicable ChatGPT data controls.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
