ChatGPT Chrome Plugin Now Reads Tabs and Fills Forms, Sparking Enterprise Data Privacy Concerns
OpenAI’s update allows the assistant to read open tabs and perform actions like form filling, raising questions about data exposure in enterprise environments.
Key points
- OpenAI announced the expanded Chrome integration on 15 October via an X thread.
- Work mode enables the assistant to fill forms, send messages and download files.
- Users can tag the @Chrome plugin to share tab content with the desktop app.
OpenAI has updated its ChatGPT extension for Google Chrome to allow the assistant to access information from open browser tabs and perform automated actions on webpages. This shift moves the tool from passive information extraction to active workflow participation within the browser environment.
Root cause
The previous integration limited the assistant to extracting text from visible tabs, which restricted its utility for tasks requiring direct interaction. The new update splits the experience between the Chrome sidebar and the full ChatGPT interface, allowing deeper context sharing. This architectural change enables the system to maintain awareness of the user’s current browsing state across multiple windows.
Attack path
In Work mode, ChatGPT can directly interact with supported webpages to execute specific tasks. The assistant can fill out forms, send messages, download files and organize tabs without manual user input for each step. One demonstration showed the system combining details from several open tabs to rank products, illustrating how it processes and acts upon cross-tab data.
Affected versions
The changes apply to the ChatGPT Chrome extension and the associated desktop or web application interfaces. Users must explicitly tag the @Chrome plugin within their ChatGPT conversations to bring existing browser tab data into the chat context. The sidebar feature also updates to support these new interaction capabilities within the browser window.
Mitigation
- Review the permissions granted to the ChatGPT Chrome extension in browser settings.
- Disable Work mode when handling sensitive accounts or confidential information.
- Monitor browser activity for unexpected form submissions or file downloads.
- Ensure organisational policies address the use of AI tools with browser access.
Why it matters
Teams usually treat browser extensions as low-risk utilities, but automation capabilities change the threat model. When an assistant can read tabs and click buttons, it gains implicit trust in the user’s active sessions. Security teams should check whether the tool accesses sensitive domains or stores session data. Guidance for this kind of event involves verifying that the assistant only interacts with approved sites and that users understand the visibility of their browsing context.
What to do and how to stay safe: OpenAI
- Audit which browser extensions have permission to read and change data on websites.
- Restrict the use of AI automation features in environments handling personal or financial data.
- Train users to recognise when an AI is acting on their behalf rather than just advising.
- Monitor logs for unusual browser activity that may indicate automated form submissions.
Step-by-step guide: Stop Sensitive Data Leaks in AI Chatbots: Priority Controls
General security guidance from the Vector Bulletin newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
When was this update announced?
OpenAI announced the changes on 15 October through a thread on X.
Can ChatGPT fill out forms in Chrome?
Yes, in Work mode the assistant can fill forms, send messages and download files.
How does the assistant access tab data?
Users can tag the @Chrome plugin in ChatGPT to share information from open tabs.
