The newest Gemini in Chrome mobile AI browser collects the most user information out of all AI browsers recently analysed by Surfshark.
It collects data such as users’ names, precise location, and browsing history. And a cybersecurity expert has drawn attention to an even more important fact – we may face a new type of surveillance when AI browsers see, track, and interact with everything we do online.
Even though agentic browsers are so powerful that they can improve our lives by taking over tedious tasks such as ordering groceries or booking a haircut, Karolis Kaciulis, leading system engineer at Surfshark, highlights the other side of the coin which is that AI browsers can see everything we humans do on the screen.
“Users’ privacy and personal information are becoming a cheap commodity,” Kaciulis says. “An agentic browser needs to see and monitor everything we do online. Your name and physical address, emails, text messages, the fact that you are stalking your ex – this is just a small part of what an agentic browser knows about you. No room for privacy left.”
As stated on the Apple App Store, Gemini in Chrome mobile integration collects your name, email and physical addresses, payment info, emails or text messages, even sensitive info such as racial or ethnic data, sexual orientation, disability, and much more extremely private data.
Most importantly, Gemini and Chrome link this data to a user.
This means that they’re not just seeing what you are doing in general on the screen – they are actually creating a specific and detailed profile of you. Why are the companies doing that? The main reason may be to provide that information to generative AI, create personalised content for a user, and boost profits.
“My key question is: do companies see users’ privacy and security as their main focus area?” asks Kaciulis. “Otherwise, what consequences are we, as users, going to face after one, five, or 10 years of feeding agentic browsers with our most private and sensitive information?”
With the spurt of agentic browsers, Perplexity has already announced a major security flaw with its Comet AI browser which potentially let hackers gain access to extremely sensitive user information such as bank account details.
Kaciulis says steps to protect your privacy and security when using an agentic browser include:
- Don’t allow it to handle your sensitive data. Never allow an agentic browser to fill out forms with highly personal information like your social security number, passwords, or bank details. Always type that information yourself.
- Choose a traditional browser. It’s convenient to ask the agentic browser to search and summarise the best Christmas time flight options for you. However, always use a standard and privacy-focused browser when you are checking your email, doing online banking, or accomplishing other tasks that require your personal information.
- Check the privacy settings. Go into your browser’s settings and look for the “Privacy” or “Data Controls” section. Look for options to “opt out” of data sharing or to delete your past conversations with the AI.
- Review the AI’s actions before you approve. Your agentic browser may try to do things on its own such as clicking a button or filling in a field. For example, when submitting a form or survey or clicking “Next” in a multi-step process. Always take a moment to look at what it’s doing before you give the final “OK”. If something looks strange or suspicious, stop the process.
- Update it. Your browser and its AI features need regular updates which often include security fixes. Make sure it’s set to update automatically so you’re always using the safest version.