Microsoft Uses AI to Combat Tech Support Scam

with No Comments

At some point, almost every internet user has the same experience. You’re going about your day when suddenly a popup appears on the screen. It says your computer is infected with a virus and requires immediate assistance from tech support. When you try to close the popup, it either doesn’t close or simply shows up again, forcing you to take somewhat drastic measures to get it to disappear.

While many people know that popups such as these are scams, some people are duped into calling the number. And, after they place the call, they’re connected to a scammer who tries to charge them for costly repairs or removal software, none of which are actually needed.

The Federal Trade Commission (FTC) decided to crackdown on these scams, and Microsoft played a big role in combating these tech support crimes.

Microsoft Digital Crimes Unit

Microsoft has its own internal Digital Crimes Unit that focuses on tracking and preventing cybercrime. The group hears over 10,000 complaints a month from people who’ve been contacted by scammers pretending to be legitimate tech support professionals, either through popup ads or telephone calls. And, since not everyone reports that they’ve been contacted by a scammer, the estimated number of people who encounter it is likely substantially higher.

Since scammers are generally very good at hiding their tracks, and many people who filed complaints had limited information to provide, Microsoft knew it had to go further if it hoped to capture the people behind the large-scale operations.

Fighting Tech Support Scammers

To begin, Microsoft created a model that searched for content that behaved a particular way. For example, popups that refreshed in mere microseconds (making it appear as though they won’t close when presented to users) were targeted by the company as a source of information. As they tracked down offending sites, they collected as much data as possible, including screenshots of the questionable content.

Since manually sorting through the data was practically impossible, Microsoft turned to machine learning, a form of AI, to help manage the information and identify patterns.

The AI was set up on Microsoft’s Azure computing platform, giving it a substantial amount of power and resources, and set loose to seek clues, like image and content similarity, to determine whether a popup was potentially relevant to their investigation. Next, they added a computer vision API from Microsoft Cognitive Services to focus on items like phone numbers to try and identify the origin of the ads.

The approach allows them to keep up with fast-moving scammers and focus man hours on finding the large-scale operations they were targeting. Finally, they used Power BI, a data visualization tool, to make the results accessible to law enforcement officials and attorneys who may be less familiar with Microsoft’s technology.

Ultimately, this created a valuable source of information for the FTC while also helping government officials learn more about how these scams affect people. And Microsoft continues to improve their processes, identifying new ways to track scammers and help keep people safe.


Similar systems are actually available to companies interested in scanning vast quantities of unstructured data, such as what is found on social media, to help identify relevant patterns that can improve their businesses. If you would like to learn more, the team at Solving IT can help. Contact us to speak with one of our knowledgeable team members today.