Google Scales Up AI-Powered Ad Enforcement
Google has announced a significant expansion of its Gemini AI models in the fight against harmful advertising on its platforms. The move comes as scammers and threat actors continue to refine their methods to slip malicious ads past existing detection mechanisms. In a newly published report, the company revealed it blocked or removed 8.3 billion ads and suspended 24.9 million advertiser accounts throughout 2025, among which 602 million ads were directly tied to scam operations.
The scale of the problem underscores just how persistent malvertising has become. Attackers routinely purchase Google Ads placements to impersonate well-known brands and services, using those placements to distribute malware, drain cryptocurrency wallets, or redirect users to phishing sites designed to harvest credentials.
How Malvertising Campaigns Operate
Modern malvertising campaigns are deceptive by design, frequently employing cloaking techniques and URL redirects to make malicious destinations look like trusted websites. In some cases, attackers spoof Google's own domains or mimic legitimate software download pages and authentication portals to lend credibility to their ads.
Recent campaigns documented by BleepingComputer illustrate the breadth of these tactics:
- Fake login pages crafted to steal Google Ads account credentials from advertisers themselves.
- Trojanized software distributed through ads impersonating popular tools such as Google Authenticator and Homebrew.
- Ads promoting counterfeit cryptocurrency platforms engineered to drain visitors' digital wallets upon interaction.
Security researchers at Malwarebytes have also documented malicious Google ads that convincingly impersonate Google.com itself, further blurring the line between legitimate and fraudulent content for everyday users.
Cybercriminals Now Wielding Generative AI
A troubling development highlighted in Google's report is that cybercriminals have begun incorporating generative AI into their own workflows. This capability allows bad actors to construct more sophisticated, large-scale deceptive ad campaigns far more rapidly than was previously possible through manual methods.
Google's VP & General Manager of Ads Privacy and Safety, Keerat Sharma, addressed this directly:
"Bad actors are using generative AI to create deceptive ads at scale, and Gemini helps us detect and block them in real time. By the end of last year, the majority of Responsive Search Ads created in Google Ads were reviewed instantly, and harmful content was blocked at submission — a capability we plan to bring to more ad formats this year."
What Makes Gemini's Approach Different
Earlier ad review systems relied heavily on keyword analysis to flag potentially harmful content, a method that sophisticated attackers could sidestep with relative ease. Google's Gemini-powered enforcement infrastructure takes a fundamentally broader approach. Rather than scanning for specific words or phrases, Gemini can simultaneously analyze billions of signals — including advertiser behavior patterns, account history, campaign structures, and declared intent — to build a comprehensive picture of whether any given ad poses a risk.
This holistic analysis is applied before ads are ever shown to users, allowing Google to intercept malicious submissions at the point of entry rather than reacting after harm has already occurred.
U.S. Enforcement Numbers and Policy Violations
In the United States specifically, Google removed 1.7 billion ads and suspended 3.3 million advertiser accounts during 2025. The top two policy violations cited were "abusing the ad network" and "misrepresentation", reflecting the prevalence of impersonation-based schemes targeting American users.
Faster Response to User Reports and Fewer False Positives
Beyond proactive blocking, Gemini has also improved Google's ability to act on user-submitted reports of suspicious ads. The AI's capacity to process and triage incoming reports has meaningfully accelerated response times compared to previous years.
Equally significant is the impact on advertiser experience. Google stated that the increased accuracy of its AI detection models has resulted in a reduction of incorrect advertiser suspensions by 80%. This improvement addresses a long-standing criticism that automated enforcement systems too often penalized legitimate businesses alongside genuinely malicious actors.
What Comes Next
Google says it plans to continue expanding Gemini's role across additional ad formats and enforcement systems throughout the coming year. The overarching goal is to move enforcement as far upstream as possible — catching harmful campaigns at the moment of submission rather than after they have had any opportunity to reach end users.
As generative AI lowers the barrier for bad actors to produce convincing, large-volume malicious ad campaigns, Google's bet on deploying its own AI infrastructure at scale represents a direct technological counter. Whether this arms race ultimately tips in favor of defenders or attackers remains an open question, but the 2025 enforcement figures suggest Gemini is already having a measurable effect on the volume and reach of malvertising across Google's network.
Source: BleepingComputer