Google Ads Uses New AI Model to Detect Fraudulent Advertisers

Google is updating how it fights fraud in the Google Ads ecosystem with a new artificial intelligence system designed to identify and block dishonest advertisers more effectively. The company published details of the model called ALF in a research paper on December 31, 2025.

What Is ALF?

ALF stands for Advertiser Large Foundation Model, a multimodal AI architecture developed by Google researchers. Unlike older systems that examined individual signals in isolation, ALF analyzes a wide range of advertiser data, including text, images, videos, account age, billing details, and historical performance, to build a detailed picture of advertiser behavior. This helps Google distinguish between legitimate accounts and those with potentially fraudulent intent.

The model’s multimodal approach makes it much more accurate at spotting scams and policy violations than previous tools, with published tests showing ALF can improve recall by more than 40 percentage points on key safety tasks.

How Does ALF Improve Fraud Detection?

According to the research, many factors, like a newly created account, ad creative that resembles a brand name, or a declined credit card, might appear safe on their own. But when combined, these signals can strongly suggest malicious intent. ALF’s ability to analyze all signals together allows the system to detect subtle patterns that would otherwise slip through traditional filters.

Google also uses a technique called inter-sample attention, which means the model doesn’t look at each advertiser in a vacuum. Instead, it compares a large batch of advertisers as a group to understand what “normal” behavior looks like.

To protect privacy, Google strips out personally identifiable information before feeding data into the model. This ensures that the AI identifies suspicious behavior patterns rather than sensitive personal details.

Google Brings More Natural Voice Conversations to AI Search

Why It Matters for Advertisers and Users

This new AI-driven approach could significantly benefit both advertisers and users:

Cleaner ad ecosystem: by catching fraud more accurately, Google can reduce wasteful or deceptive ad spend, helping genuine advertisers compete more fairly;
Improved trust: users are less likely to encounter misleading or harmful ads if Google’s enforcement systems can spot fraud earlier and with greater precision;
Fewer false bans: with higher precision on some policies, ALF aims to avoid wrongly penalizing legitimate advertisers – a common concern with automated enforcement;
Preparation for evolving threats: as fraudsters adopt more sophisticated tactics, a model that understands patterns across multiple data types is a more reliable option.

Google’s implementation of ALF can become a major step in keeping digital advertising safer and more trustworthy for marketers and users.

More related news:

What do you think?