ChatGPT Can Now Do More in Chrome, but There’s a Security Trade-Off

ChatGPT can now do more than tell you what to do online; it can take over parts of the work itself.

OpenAI has expanded ChatGPT’s integration with Google Chrome, allowing the assistant to use information from open tabs and, through Work mode, interact with webpages to perform tasks such as filling out forms, sending messages, downloading files, and organizing tabs.

Announced on Oct. 15 through an X thread, the update brings ChatGPT closer to handling entire browser-based workflows rather than simply answering questions. But the added convenience also raises questions about how much access users and businesses should give an AI assistant, particularly when working with sensitive information or accounts.

New abilities coming to ChatGPT’s Chrome extension

ChatGPT’s earlier Chrome integration primarily allowed the assistant to extract information from open tabs, limiting its usefulness for tasks that required direct interaction with webpages.

The company has now updated the extension to let ChatGPT do more with users’ existing browser sessions, splitting the experience between its sidebar and the full ChatGPT interface in the desktop app or web version.

The extension now brings a fuller ChatGPT experience into Chrome’s sidebar. This lets users continue existing conversations or start new tasks alongside the webpage they are viewing, without switching to a separate ChatGPT window.

The more significant change is that users can now bring their existing Chrome tabs into ChatGPT conversations on the web or desktop app. Instead of manually copying links or content, they can tag the @Chrome plugin in ChatGPT to work with information across tabs.

In Work mode, ChatGPT can interact with Chrome tabs to fill out forms, send messages, download files, and group or organize tabs. In one example, OpenAI demonstrated a workflow in which ChatGPT combined details from several open Chrome tabs to rank sofas for a user’s apartment.

The sidebar and tab-sharing features help ChatGPT access browser context, while Work mode extends that functionality to supported actions on webpages.

Must-read security coverage

Privacy issues may hamper availability

OpenAI says the feature is available to users across all ChatGPT plans, including Free, while PCMag reports support for Chromium-based browsers. That means users of Chrome, Edge, Brave, Opera, and Vivaldi can access the integration.

However, this convenience comes with privacy considerations, especially because ChatGPT can now interact with websites where users are already signed in.

OpenAI says the extension can read and change data, manage downloads in bulk, and organize tabs. Such access can expose sensitive information if users allow the assistant to work with confidential pages or accounts.

There is also the risk of malicious instructions embedded in webpage content. A compromised or deceptive page could contain instructions designed to manipulate an AI assistant into taking unintended actions or exposing information. As such, users should treat webpage content as untrusted and review the sites they ask ChatGPT to use.

Takeaways for enterprises

OpenAI’s expanded Chrome capabilities are the kind of productivity features enterprises are likely to appreciate, particularly for employees who spend much of their time researching information, working across multiple websites, and handling repetitive browser-based tasks.

By letting ChatGPT use existing browser context and carry out supported actions, the update could reduce manual work and help employees complete tasks in fewer steps.

However, businesses considering the integration should evaluate more than its productivity benefits. Organizations handling customer records, financial information, or internal documents need to understand what data ChatGPT can access, which websites employees can authorize, and what safeguards apply when the assistant performs actions.

The risks also extend beyond information access. A mistake or malicious instruction embedded in webpage content could potentially lead an AI assistant to perform unintended actions, particularly when working with authenticated accounts.

Enterprises should therefore establish clear usage policies, limit access to sensitive systems where appropriate, and require human review for consequential actions such as sending information, modifying records, or submitting forms.

The goal should not be to automate every possible task, but to identify workflows where ChatGPT can save time without receiving more access or authority than necessary.

Other news: The US Department of Homeland Security has ordered federal civilian agencies to remove DeepSeek products from their IT systems, citing concerns over sensitive data exposure and potential security vulnerabilities.

About Author

AndyC

Andy Curtis is an award-winning security consultant, researcher and public speaker. He has been working in the computer security industry since the early 1990s, having been employed by state and federal government, leading healthcare and banking providers across three continents. He has given talks about computer security for some of the world’s largest companies, worked with law enforcement agencies on investigations into hacking groups, and is a regular voice on TV and radio explaining IT security threats.

See author's posts