Google's AI blocked nearly half a billion harmful ads in India last year. Here's how

Google has taken down nearly half a billion harmful advertisements in India over the past year, underscoring both the scale of online abuse and the growing role of artificial intelligence in policing it.

In its 2025 Ads Safety Report, the company said it blocked or removed 483.7 million ads in India for violating its policies, while suspending 1.7 million advertiser accounts linked to repeated or serious breaches. The actions were driven largely by its AI systems, including Gemini.

AI moves from reaction to prevention

Google said its enforcement systems are now shifting from reactive moderation to early detection, powered by AI models that analyse intent and behavioural signals.

“Bad actors are using generative AI to create deceptive ads at scale, and Gemini helps us detect and block them in real time,” the company said, adding that more than 99% of violating ads globally were blocked at an early stage.

Globally, Google said it removed or blocked over 8.3 billion ads in 2025.

Financial scams, impersonation remain key risks

In India, the bulk of violations fell into a handful of recurring categories. These included misleading financial services ads, trademark and copyright infringements, and misuse of personalisation tools for targeted manipulation.

The company flagged financial fraud and impersonation as particularly concerning, noting that bad actors continue to evolve tactics to bypass safeguards, often mimicking trusted brands or exploiting user data for targeted scams.

Crackdown on repeat offenders

Apart from removing individual ads, Google said suspending advertiser accounts is critical to stopping repeat abuse. Google shut down 1.7 million advertiser accounts in India that repeatedly violated its policies or tried to bypass its enforcement systems.

The company said its teams have long used advanced AI to identify and stop scammers and Gemini has taken that work even further. 

"Our models analyse hundreds of billions of signals, including account age, behavioural cues and campaign patterns, to stop threats before they reach people," the company said. 

"Unlike earlier keyword-based systems, our latest models better understand intent, helping us spot malicious content and preemptively block it, even when it’s designed to evade detection," it added.