Google Uses Gemini AI for Millisecond Detection, Blocking 99% of Scam Ads

Online fraud rings are using automated tools at scale to place bogus investment, impersonation and phishing ads, making it difficult for traditional human review to respond in real time. Google has therefore deployed Gemini AI to analyze signals including ad content, payment information and account behavior, using millisecond-level assessments to reduce the risk of harmful ads reaching users.

Google’s latest annual ads safety report said Gemini AI can now intercept 99% of scam ads that violate its policies. In 2025, Google removed more than 8.3 billion ads worldwide and suspended over 24 million advertiser accounts. In Taiwan, it also removed 3,564 URLs linked to fraudulent activity.