Google has quietly rolled out a new AI model inside its Google Ads system to fight fraud and policy violations.
The company revealed details in a research paper published on December 31, 2025.
The new system has already gone live. Google says it improves fraud detection by more than 40 percentage points and achieves 99.8% precision on key ad policies.
This marks a major upgrade in how Google protects advertisers and users from scams.
How the New AI Model Works
The AI system is called ALF, short for Advertiser Large Foundation Model. It is a multimodal AI model, meaning it analyzes many data types together instead of in isolation.
Key Specifications of ALF:
- Analyzes text, images, and video ads
- Reviews account age and billing details
- Studies historical performance data
- Evaluates landing page content
- Compares patterns across large advertiser groups
- Uses inter-sample attention to spot unusual behavior
- Removes all personal identifiable information (PII) before analysis
Google explains that single signals may look harmless on their own. But when combined, they can reveal fraud. For example, a new account using famous brand ads and failed payments may indicate a scam.
ALF also solves problems older systems struggled with. Advertisers can upload thousands of assets. Fraudsters often hide bad ads among many safe ones. The new model can detect these patterns more accurately.
Also read about: Google Ads Statistics 2026
Accuracy, Privacy, and Real-World Impact
Google reports that ALF performs better than its previous production systems. It improves both recall and precision at the same time.
While the model takes slightly longer to respond, Google says the delay is small and acceptable. ALF already handles millions of requests daily.
Key Benefits:
- Boosts fraud detection accuracy
- Reduces false positives
- Improves trust for honest advertisers
- Protects users from malicious ads
Google confirms that ALF is now part of the Google Ads Safety system. It is not yet used in Search or Business Profiles.
Future updates may include tracking behavior changes over time and helping with ad optimization.
The rollout shows how Google Ads fraud detection is becoming more advanced as AI models grow stronger and more reliable.
More News To Read:
