Our security groups work across the clock to cease dangerous actors that use more and more refined, malicious advertisements. In 2025, Gemini-powered instruments dramatically improved our skill to detect and cease dangerous advertisements: Our programs caught over 99% of policy-violating advertisements earlier than they ever served, and we’re persevering with to evolve our defenses to remain forward of even essentially the most superior schemes.
Blocking advertisements with better precision
Our groups have lengthy used superior AI to determine and cease scammers, and Gemini takes that work even additional. Our fashions analyze a whole lot of billions of alerts — together with account age, behavioral cues and marketing campaign patterns — to cease threats earlier than they attain folks. In contrast to earlier keyword-based programs, our newest fashions higher perceive intent, serving to us spot malicious content material and preemptively block it, even when it’s designed to evade detection.
This proactive method helped us defend in opposition to dangerous actors. In 2025, we blocked or eliminated over 8.3 billion advertisements and suspended 24.9 million accounts, together with 602 million advertisements and 4 million accounts related to scams.
Whereas know-how catches the overwhelming majority of coverage violations, our expansive advertiser verification program offers one other important layer of prevention. After we validate identities and cease dangerous actors earlier than they enter our system, we construct belief within the advertisements folks see and the companies behind them.
Working quicker than scammers
Dangerous actors are utilizing generative AI to create misleading advertisements at scale, and Gemini helps us detect and block them in actual time. By the top of final yr, the vast majority of Responsive Search Adverts created in Google Adverts have been reviewed immediately, and dangerous content material was blocked at submission — a functionality we plan to carry to extra advert codecs this yr.
Gemini additionally allows us to course of consumer suggestions extra effectively, which helped our groups to take motion on greater than 4 occasions as many consumer studies in 2025 than within the yr prior. Which means when a risk does slip via, we will neutralize it shortly. This velocity lets our security specialists deal with the advanced work that requires human judgment.
Stopping threats whereas supporting companies
By analyzing past pictures and textual content patterns, Gemini can higher distinguish between a reputable provide and an intricate lure — a stage of nuance that helped us reduce incorrect advertiser suspensions by 80% final yr. This accuracy permits us to deal with two issues in tandem: prioritizing the removing of dangerous content material whereas serving to trustworthy companies hold their advertisements operating.
As threats evolve, Gemini retains our defenses one step forward. Be taught extra within the 2025 Ads Safety Report.
