Google has published its 2025 Ads Safety Report, revealing how Gemini-powered tools are reshaping its defences against malicious advertising — blocking over 8.3 billion ads and suspending 24.9 million accounts over the course of the year.

Gemini goes beyond keywords

Earlier ad safety systems relied heavily on keyword-based detection. Google says its Gemini models now analyse hundreds of billions of signals — including account age, behavioural cues, and campaign patterns — to identify malicious intent before ads reach users. The shift enables more proactive blocking of content designed to evade detection, including ads generated by bad actors using generative AI at scale.

Among the accounts suspended, 4 million were linked to scams, alongside 602 million scam-related ads blocked. Google says its systems caught over 99% of policy-violating ads before they were ever served.

Real-time review and faster response

By end of 2025, the majority of Responsive Search Ads created in Google Ads were being reviewed instantly, with harmful content blocked at submission. Google said it plans to extend this capability to more ad formats in 2026.

Gemini also accelerated the processing of user-reported violations. Google said its teams acted on more than four times as many user reports in 2025 compared to the prior year — allowing human safety experts to focus on complex cases requiring judgment rather than volume.

Fewer false positives, fairer enforcement

One notable outcome from Gemini’s deeper contextual analysis is a reduction in erroneous advertiser suspensions. Google reported an 80% drop in incorrect suspensions, attributing it to the model’s improved ability to distinguish between a legitimate offer and a sophisticated scam — a nuance that earlier pattern-matching systems struggled with.

The report positions Gemini not just as a tool for enforcement, but as a mechanism for building trust in the broader ads ecosystem — validating advertiser identities upstream and reducing the load on post-serve remediation.