The bots have arrived. And they like to click things.
According to new research from DoubleVerify, which they compiled while building new tools to help advertisers see when AI scrapers or agents interact with their ads, AI bots accounted for up to 15% of all ad clicks in DV’s tests of “unprotected media.” In some client cases, DV found that “an ad click was 10X more likely to be from an AI bot than from a human.”
In the words of Scooby Doo: ruh-oh.
DV notes that bots and scrapers like these aren’t necessarily or inherently malicious. In fact, they typically fall under General Invalid Traffic (GIVT) — activity that should be removed from reporting. But, of course, to remove it, you first have to know when bots are showing up.
Why This Matters:
This isn’t your grandparents’ GIVT. This is AI GIVT, and the motivations are different.
DV says many of these bots click on ads not to commit fraud, of course, but because they’re learning. AI agents often break tasks into smaller subtasks — like scraping, testing, or simulating a user journey — and those subtasks can involve clicking. A lot. They click to learn, with ads collateral damage.
In fact, DV’s research, which included looking at AI from companies like Perplexity and OpenAI, shows that some bots run 24/7, inflating engagement metrics via repeated clicking that goes nowhere and doesn’t convert. DV also found that AI agents are becoming more evasive, often manipulating browser credentials or intentionally mimicking human engagement to avoid detection.
In several tests, the company identified clusters of these evasive scrapers responsible for more than 10 million impressions per day. And in one case, a single bot “hopped” across 40 different local news sites in a single day, each in a different U.S. city — a pattern that could obviously distort performance data without deeper inspection or visibility.
For brands, this adds up to a familiar problem with a new twist: media waste created not by fraudsters, but by autonomous systems that behave like people but aren’t.
Experts React:
On LinkedIn, Gilit Saporta, who heads up DoubleVerify’s fraud work, said:
“Bots that don’t learn quickly enough will click again, and again, and again. They’ll fail in their agentic task… they’ll also skew metrics, cause confusion and… generate waste.”
Our Take:
We’re entering a world where much of the content on the internet will be AI-generated, not human-generated, especially as the tools get better and cheaper. In the same vein, it’s not hard to imagine a future where a meaningful share of ad clicks also comes from AI agents and bots. If that’s where we’re headed, advertisers, publishers, and everyone in adtech need to be prepared.
AdTechRadar is owned by Chris Harihar, who leads PR at Mod Op. DoubleVerify is a Mod Op client.
