Many people use ad blockers to avoid annoying or inappropriate ads while browsing the internet.
But a new study suggests that these tools might not be protecting users as well as they think—and could even be making things worse.
Researchers from the NYU Tandon School of Engineering looked at over 1,200 ads in the U.S. and Germany.
They focused on Adblock Plus, one of the most popular ad-blocking tools, which has a feature called “Acceptable Ads.”
This feature allows certain ads to show if they meet specific guidelines meant to make them less disruptive. The idea is to find a balance between supporting websites that rely on ads and giving users a better browsing experience.
But the study found that users who had the Acceptable Ads feature turned on actually saw 13.6% more problematic ads than people who didn’t use any ad blockers at all.
Problematic ads include things like inappropriate content for kids, misleading health or financial claims, fake countdown timers, and even ads that trick people into clicking on scams.
To identify these bad ads, the researchers created an automated system using AI. They also came up with a clear definition of what makes an ad “problematic,” based on industry rules, government guidelines, and feedback from users. Their AI tool, powered by OpenAI’s GPT-4o-mini model, agreed with expert reviewers nearly 80% of the time.
The findings were especially troubling when it came to ads seen by younger users. Around 10% of the ads shown to underage users broke rules designed to protect children from harmful or inappropriate content.
What’s even more concerning is that ad exchanges—the systems that deliver ads to websites—seem to treat users with ad blockers differently. The researchers found that some ad platforms actually sent more bad ads to people who had Acceptable Ads enabled. In contrast, newly added ad partners tended to behave better.
This pattern raises a worrying possibility: that ad platforms might be able to detect who’s using an ad blocker and deliberately send them lower-quality or even riskier ads. The study suggests this could be a new kind of digital “fingerprinting,” where people become easier to track online simply because they’re trying to protect their privacy.
Overall, the study shows that tools like Acceptable Ads may not live up to their name. While they aim to improve the browsing experience, they could unintentionally open the door to more harmful content—especially for privacy-conscious users.
The research will be presented at the Privacy Enhancing Technologies Symposium on July 15, 2025.
