AI detects fraudulent online shops

Your Link is Almost Ready Scroll Down After Loading...

A browser plug-in, which will be available soon, will automatically detect fraudulent online shops (fake shops). To do this, the AI ​​analyzes 21,000 factors.

There are numerous fraudulent online shops on the Internet where buyers receive no goods and lose their money. According to the Austrian platform Watchlist Internet, 3,182 of the 16,500 reported Internet fraud cases were in so-called fake shops in 2020. In Germany, too, fake shops have been a problem that has persisted for years.

A team from the Austrian Institute for Applied Telecommunications (ÖIAT), in cooperation with the Austrian Institute of Technology (AIT) and the SME research network Austrian Cooperative Research (ACR), has developed an artificial intelligence (KI) that automatically warns buyers of fraudulent online shops. The development was financed by the research promotion company FFG as part of Kira's security research program.

Detect fake shops automatically

Users can often identify fraudulent online shops based on features such as strange domain names, few reviews, low prices, and prepayment as the only payment method. However, this is not always possible in professional fake shops. The developers, therefore, identified more than 21,000 factors that are indications of a fake shop, including the code, the language, and the page structure. When entering the page, the AI ​​analyzes these factors automatically.

"We asked ourselves whether the similarities of the fake shops that we see from the outside can also be recognized by the machine," explains project manager Louise Beltzung.

7,700 fraudulent online shops as a database

As a database, the FIAT compiled a list of 7,700 fake shops, which the AI ​​analyzed to identify similarities. On the basis of the training data, the software now carries out an analysis in real-time when an unknown page is entered, which examines the typical characteristics of fake shops. The recognition rate is already 97 percent but should be improved. “Quality assurance is one of the greatest challenges in AI solutions. This applies above all to the database, ”explains Beltzung.

Despite the already excellent results, the researchers do not want to leave the final decision to the software in uncertain cases. If the AI ​​is not completely sure, a guide is displayed to the user, which he can use to examine the alleged fake shop himself.

The Fake Shop Detector will be available in the form of a free browser plug-in in a few weeks.