Google is using Gemini AI to fight ad fraud

Ad fraud has long been a silent tax on digital campaigns, siphoning away ad dollars without a trace.

Now, Google says it’s found a way to fight back by using its Gemini large language models (LLMs) to sniff out deceptive ad activity across the web and mobile apps.

This article explores how Google is applying Gemini to detect invalid traffic, what that means for mobile ad quality, and how marketers can use this shift to their advantage.

Short on time?

Here’s a table of contents for quick access:

Google is using Gemini AI to fight ad fraud

What is Google doing with Gemini AI?

Since late 2023, Google has been running a quiet pilot project using Gemini-powered models to analyze web and mobile ad environments. The models are trained to behave like real users, clicking, scrolling, navigating apps, and capturing screenshots so they can detect when something’s not right.

The AI is designed to flag “deceptive or disruptive” ad tactics such as:

Hidden ads that are technically visible but can’t be seen by a real user
Ads that force user interaction to proceed
Accidental clicks and non-human interactions

In a 10-month pilot from December 2023 to October 2024, Google saw a 40% reduction in mobile invalid traffic (IVT) tied to these disruptive ad formats. While the process isn’t fully automated yet, it’s already speeding up policy enforcement and human review.

Why invalid traffic is a billion-dollar problem

Invalid traffic includes clicks and impressions from bots, accidental taps, or even malware-induced behavior. According to Pixalate, IVT rates in Q1 2025 were 18% on the web and a staggering 31% on mobile apps. That signals a serious efficiency leak for digital advertisers.

“If you [as an advertiser] spend $100,000, and $10,000 goes to invalid traffic, that’s $10,000 wasted that you could have spent on actually acquiring users,” Per Bjorke, Google’s director of product management for ad traffic quality, said. “It’s also a problem for good publishers because any dollar paid to bad actors is a dollar that should have gone to a good publisher. It puts the whole advertising ecosystem at risk.”

The risk isn’t just financial. Bjorke pointed to a case where a pop-up ad prevented a user from dialing 911, and another where malware-infected apps generated 1.5 billion fraudulent bid requests per day.

This year alone, Google has:

Removed 352 Android apps tied to an ad fraud ring
Deleted nearly 200 more in a separate operation
Filed lawsuits against China-based hackers tied to malware-driven ad schemes

What marketers should know

Marketers don’t need to be ad fraud experts to stay protected, but they do need to know what’s changing behind the scenes. Here’s what matters:

1. AI is changing ad enforcement

Gemini isn’t just scanning code. It’s interacting with apps like a user, revealing bad behavior that traditional analysis might miss. AI-driven enforcement is becoming faster, more precise, and more scalable.

What to do: Ask your media partners how they’re incorporating AI-based traffic filtering and whether they meet Google’s IVT standards.

2. Not all impressions are created equal

With IVT rates reaching up to 31% on mobile, marketers should question the true value of programmatic impressions, especially in app-heavy campaigns.

What to do: Push for transparency in inventory sourcing, and consider working with platforms that certify traffic quality through third-party verification.

3. Use human plus machine review as your benchmark

Google’s approach shows that AI can detect IVT efficiently, but human oversight is still needed to interpret nuance and enforce policies correctly.

What to do: Don’t fully automate ad fraud detection on your end. Blend AI tools with human QA teams to validate suspicious patterns.

4. Ad fraud is a moving target

As Bjorke puts it, “This is always an adversarial game.” Bad actors evolve as quickly as platforms do. AI tools like Gemini help advertisers stay one step ahead by adapting in real time.

What to do: Stay updated on platform security advancements and consider auditing your campaigns quarterly for traffic anomalies.

Google’s Gemini rollout signals a broader shift in how ad platforms are using AI to protect ecosystem trust. That’s good news for performance marketers, but only if you hold partners accountable and insist on verified quality.

The war against invalid traffic isn’t over, but marketers now have a smarter ally on their side.

This article is created by humans with AI assistance, powered by ContentGrow. Ready to explore full-service content solutions starting at $2,000/month? Book a discovery call today.