Google is now targeting bad ads for bad behavior

Google said Thursday it will block 8.3 billion ads worldwide in 2025 – from 5.1 billion last year. But the company suspended fewer ad accounts than it said it would, raising questions about how it manages its platform.

The search giant attributed the discrepancy to its use of AI, particularly its Gemini models – Google’s family of AI systems – which Google says allows it to detect and block infringing ads earlier and with greater accuracy. Its AI-powered system caught more than 99% of ads last year before they were shown to users, the company said.

All of these findings come from Google’s 2025 Ads Safety Report and together they show a significant improvement in ad targeting. While problematic ads are being blocked, fewer ad accounts are being blocked – reflecting a shift from blocking bad actors to blocking ads on their own.

Google says the rise in banned ads also reflects the increasing use of AI by agents to create fraudulent content at scale, with its Gemini models helping to identify patterns in large campaigns and block them earlier.

The update also reflects Google’s big push to integrate its Gemini models very deep in its main features and construction, including advertisingwhere the company is increasingly using AI to create campaigns, detect breaches, and respond to threats in real time.

Of the banned ads and suspended accounts, 602 million ads and 4 million ad accounts were linked to fraud, the company said.

Google has removed more than 1.7 billion ads and suspended 3.3 million ad accounts in the US by 2025, with cyberbullying, misrepresentation, and sexual harassment among the violations. In India, Google’s largest market with users, it closed 483.7 million ads – almost double last year – although account suspensions fell to 1.7 million from 2.9 million, with trademarks, financial services, and copyright issues among the biggest violations.

At a briefing, Keerat Sharma, Google’s Senior Vice President of Privacy and Security, told reporters that the company has shifted to AI-driven security “on a larger scale, on a production scale, rather than using a more subtle tool, such as an ad blocker.” He added that the system has helped reduce wrongful terminations by 80% annually.

Google’s security measures, including advertiser verification (a process that requires businesses to verify their identity before advertising), are designed to prevent bad actors from creating accounts, Sharma said, adding that this has helped reduce suspensions.

Those numbers, Sharma said, are likely to fluctuate over time as Google rolls out new defenses and attackers, with the company aiming to stop malicious ads as soon as possible.