What Agentic Browsers Reveal About Online Privacy

TL;DR Summary

• The Challenge: Privacy claims often don't match reality. Some sites start profiling before consent.
• The Tool: An agentic browser that reads policies, detects live trackers, and flags contradictions.
• The Outcome: Clear evidence of premature tracking and policy violations on 5 of 6 news sites.
• Impact: A valuable tool for educators, researchers, and privacy-conscious users.
• Key Finding: 2 out of 3 parent companies profiled users before consent.

Project Snapshot

This project set out to answer a vital digital literacy question:

Do websites honour their own privacy commitments?

Using an AI-powered agentic browser, we conducted a real-world audit of six popular Dutch news sites. This browser didn't just display pages-it read policy texts, detected trackers, and explained everything in plain English. It was a browser, watchdog, and researcher in one.

The results revealed significant gaps between stated policies and actual practice-especially in commercial media.

For full details, see our article: Privacy Promises vs. Reality: A Case Study of Six Dutch News Sites.

⚠️ The Challenge & Context

The Illusion of Consent

Cookie banners and “Reject All” buttons give users a sense of control. But behind the scenes, tracking scripts often run before any choice is made. With regulations like the GDPR in force, that's a problem-not just for users, but for educators and watchdogs trying to hold platforms accountable.

Why Traditional Tools Weren't Enough

Ad blockers and privacy extensions are great at hiding ads, but they lack context. They don't detect when trackers start, or compare policy with behaviour. We needed something more dynamic-an agent with awareness.

Our Approach

A Human-AI Collaboration

We used an agentic browser-a new kind of tool that:

• Reads privacy policies
• Monitors real-time tracking scripts
• Flags mismatches between policy and behaviour
• Explains it all in clear, non-technical language

This made it perfect for education, research, and digital audits.

Key Technologies Used

• Agentic browser (Comet)
• AI-based policy summarisation
• Tracker detection and fingerprinting
• Automated opt-out testing

Outcomes & Visuals

Tracking Scores

💡 Key Insights

• Tracking begins before consent-often instantly.
• Users are profiled even without logging in.
• Social media platforms receive data regardless of interaction.
• “Reject All” is often misleading-hidden behind layers or non-functional.

Before vs. After

Discussion & Future Directions

Lessons Learnt

• Privacy banners can mislead users into a false sense of control.
• Some platforms intentionally obfuscate opt-out settings.
• Even compliant-looking sites often fail ethical standards.

Continuous Improvement

Future agentic browsers could support:

• GDPR/CCPA compliance reporting
• Education dashboards for schools and universities
• Integration into digital rights toolkits

From Case Study to Capability Building

This project reflects our mission at Symbio6:

To make AI tools understandable and useful-especially for those navigating digital privacy and ethics.

Whether you're:

• An educator designing an AI or media literacy module
• A privacy advocate seeking more clarity
• A team exploring responsible AI adoption

We support you with:

• AI literacy workshops tailored to your field
• Privacy training rooted in real-world tools
• Expert coaching to help your team apply AI with confidence

“Privacy isn't just a checkbox-it's a learning curve. The right tools help teams think, not just click.”

Curious what this could look like for your team?

Let's talk. You can start by exploring our AI training options or booking a short intro call.