AI Enhances Ad Fraud Detection
Google suspended 39.2 million advertiser accounts in 2024. That’s more than three times the number from the previous year. The company says AI-powered tools, especially large language models (LLMs), were key in identifying fraudulent activity.
These tools look for signs like fake business identities and invalid payment information. Google claims the suspension of a bad account before they have a chance to run any ads.
Safety Upgrades with Human Oversight
Last year, Google introduced over 50 LLM upgrades across its platforms. These upgrades aimed to make ad safety more effective.
“AI models are vital, but human experts still play a big role,” said Alex Rodriguez, Google’s General Manager for Ads Safety. A team of over 100 specialists, including those from Ads Safety, Trust & Safety, and DeepMind, collaborated on this project.
Deepfake Ad Scams Under Fire
This team focused heavily on deepfake ad scams that use public figure impersonations. As a result, Google introduced 30 new policy changes and additional technical defenses.
These changes led to the suspension of 700,000 accounts and a 90% drop in deepfake ad reports.
Billions of Ads and Pages Removed
In total, Google blocked 5.1 billion ads and removed 1.3 billion pages in 2024. That’s slightly lower than 2023’s numbers, which saw 5.5 billion ads blocked and action taken against 2.1 billion publisher pages.
The company also restricted 9.1 billion ads, further reducing exposure to harmful content.
Regional Highlights: U.S. and India Lead Suspensions
The U.S. had the highest number of account suspensions. India followed, with 2.9 million suspended accounts and 247.4 million ads removed. Major violations included financial fraud, trademark misuse, and gambling promotions.
Globally, 5 million accounts were suspended due to scams. Nearly half a billion scam-related ads were taken down.
Election Ads and Transparency Efforts
With elections happening worldwide, Google verified 8,900 new election advertisers in 2024. It also removed 10.7 million election-related ads. However, Google says these were a small share of overall ad activity.
To address fairness concerns, Google improved its appeals process. Human reviewers now assess suspensions more carefully. “We also improved how we explain our decisions,” Rodriguez added.
Proactive Prevention is Paying Off
Google believes the suspension of the account for harmful ads shows early detection is working. Fewer harmful ads are created because bad actors are stopped earlier.
The tech giant says it will continue refining both AI tools and human-led safety processes in 2025.
Check out other related news:
