Osegi
Security·2 min read

Google deceptive software ads: study finds 100 million views

Researchers built an AI tool called AdLens and found thousands of deceptive software ads in Google's own archive, one of them live for over two years.

A laptop screen showing a fake red virus warning pop-up in a browser search results page.
Image: Digital Trends

One ad promised it could work out anyone's GPS location from their phone number. It ran for more than two years and was shown 14.5 million times. It is one of the deceptive Google software ads that researchers from NYU Tandon and Radboud University found in Google's own public archive.

The team built an AI tool called AdLens and used it on 188,834 software-related ads collected in March 2026 from Google's Ads Transparency Center. According to the study, tens of thousands of deceptive ads got past Google's filters and were shown more than 100 million times in Europe.

The researchers sorted the bad ads into three groups. 238 were scareware, meaning fake virus warnings, with over 16 million views. 3,346 made false claims, such as being able to recover old deleted photos, and were seen 89.9 million times. Another 258 hid who the advertiser was behind a button labelled something like "Continue" or "Access Now".

"Google have systems in place to screen adverts but we found thousands of deceptive adverts that still got to users, in some cases for years," said lead author Ritik Roongta.

What happened when the researchers reported the ads?

The team flagged 24 ads through Google's public reporting tool. Google answered "no violation" on 15 and removed 6. In one case, Google took down an ad from a domain tied to the TamperedChef malware campaign, yet 41 other ads pointing to the same site stayed live. Removing one ad, Roongta said, didn't necessarily stop dozens of others promoting the same threat.

How does this square with Google's numbers?

Google's 2025 Ads Safety Report says 8.3 billion ads were blocked or removed, including 602 million tied to scams, and that its systems caught over 99% of policy-breaking ads before they appeared. Those figures and the study measure different things, and both can be true at once.

There is a caveat. Impressions show how many people saw an ad, not whether anyone was harmed. Regulators are paying attention, though. In July, the UK's Ofcom recommended rules under the Online Safety Act that would make platforms like Google ban fraudulent advertisers, with fines of up to 10% of global revenue. "We shouldn't be relying entirely on the advertising platforms to tell us whether their protections are working," Roongta said.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:google adsadlensscarewaremalvertisingnyu tandonradboud university

More from this day

See all →

More in: Security

See all →