← Latest papers
💻 computer science

When the Abyss Looks Back: Unveiling Evolving Dark Patterns in Cookie Consent Banners

This paper introduces UMBRA, a novel detection system that identifies both established and nine newly evolved dark patterns in cookie consent banners, revealing through a large-scale study of 14,000 websites that these subtle, legally ambiguous manipulations systematically undermine user autonomy, obstruct consent revocation, and amplify privacy and security risks.

Original authors: Nivedita Singh, Seyoung Jin, Hyoungshick Kim

Published 2026-03-24
📖 5 min read🧠 Deep dive

Original authors: Nivedita Singh, Seyoung Jin, Hyoungshick Kim

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you walk into a store, and before you can even look at the products, a salesperson hands you a contract. They say, "Sign here to let us track your every move, sell your data to strangers, and follow you home."

You want to say "No," but the "No" button is tiny, grayed out, and hidden behind three layers of menus. The "Yes" button is huge, neon green, and flashing. Or worse, they tell you, "You can say no, but it will cost you $50," or "You can say no, but we'll still follow you anyway."

This is exactly what happens every time you visit a website today. The paper "When the Abyss Looks Back: Unveiling Evolving Dark Patterns in Cookie Consent Banners" is a detective story about how websites are tricking us into giving up our privacy, even when we try to say no.

Here is the story of their investigation, told in simple terms.

1. The Problem: The "Trick-or-Treat" Banners

Years ago, laws like the GDPR (in Europe) and CCPA (in California) said: "Websites must ask your permission before tracking you, and you must be able to say 'No' easily."

Websites complied, but they didn't comply honestly. They started using Dark Patterns. Think of these as "digital traps."

  • The Old Traps: Making the "Accept" button huge and the "Reject" button invisible.
  • The New Traps (The Evolved Ones): The websites got smarter. They didn't just hide the "No" button; they made the whole process of saying "No" a nightmare.
    • The Pay-to-Opt-Out: "You can stop us from tracking you, but you have to pay us a subscription fee." (Like a toll booth for your privacy).
    • The Fake Out: You click "Reject," but the website secretly installs trackers anyway, pretending you said "Yes."
    • The Maze: Saying "Yes" takes one click. Saying "No" requires clicking through five different menus, reading a 20-page legal document, and confirming three times.

2. The Solution: Meet UMBRA

The researchers built a robot detective named UMBRA (which stands for a system that doesn't care which consent tool the website uses).

Imagine UMBRA as a super-observant security guard who visits 14,000 different websites. UMBRA doesn't just look at the banner; it acts like a human user.

  1. It reads the text to see if they are lying about what cookies are.
  2. It clicks the buttons to see how many steps it takes to say "No."
  3. It checks the "backpack" (the browser's cookies) to see if the website put anything in there before you even agreed, or if it kept putting things in there after you said "No."

3. The Big Discovery: The "Abyss" is Deep

When UMBRA went on its tour, it found some scary things:

  • The "Pay-to-Opt-Out" is Real: Some websites literally ask you to pay money to stop them from tracking you. This is illegal in many places, but it's happening.
  • The "Fake Reject": On many sites, when you click "Reject," the website ignores you. It's like a bouncer at a club who says, "Okay, you can't come in," but then secretly lets you in anyway.
  • The Security Danger: This isn't just about privacy; it's about safety. Because these websites are forcing you to accept tracking, they are loading a lot of sketchy third-party scripts. The researchers found that sites using these dark patterns had 25% more cookies than normal.
    • The Analogy: Imagine your house (your computer). A normal website asks, "Can I leave a security camera in your living room?" If you say no, they leave. A dark-pattern website says, "Okay, you said no," but then they break a window, leave a camera, and leave the door unlocked. This makes your house vulnerable to hackers (XSS and CSRF attacks).

4. The Regional Differences

The researchers checked websites in the EU, the US, and the rest of the world.

  • The EU: Surprisingly, even with strict laws, the "tricky" patterns are everywhere. The "Pay-to-Opt-Out" is more common here than in the US, likely because companies are trying to find loopholes in the strict rules.
  • The US: The "Pay-to-Opt-Out" is rare, but the "Fake Reject" and "Hard to Opt-Out" patterns are just as common.

5. Why This Matters

The paper concludes that we are losing our autonomy.

  • The "System 1" Trap: Websites are designed to exploit our "System 1" thinking (fast, automatic, lazy thinking). They know we just want to get to the article or video, so they make "Yes" the path of least resistance.
  • The Security Risk: It's not just that they are stealing your data; it's that their sloppy, manipulative code is making your device less secure against actual hackers.

The Bottom Line

The authors argue that the "Cookie Consent" banner has become a theater of deception. It looks like you are making a choice, but the stage is rigged.

The Takeaway: Just because a website looks like it's asking for your permission doesn't mean it's respecting your choice. The "No" button is often a trap, and clicking it might not even stop the tracking. The researchers hope that by exposing these tricks (like the "Pay-to-Opt-Out" or the "Fake Reject"), regulators and users can finally stop the game and demand real, honest consent.

In short: The websites are playing a rigged game where the "No" button is a mirage. UMBRA is the map that shows us where the real exits are.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →