Algorithm recognizes certain patterns that make up fake profiles!

AI gets on the trail of dating fraudsters

A new software with artificial intelligence (AI) exposes fraudsters on dating portals. The program works with a hit rate of 93 percent. This approach was developed King's College London in collaboration with colleagues at the University of Bristol , Boston University and the University of Melbourne

Very current problem

“Catfishing is a real problem. According to studies, around ten percent of the profiles on free dating sites are fake and another large proportion are at least embellished. The goals behind the fakes are primarily monetary - for example, amounts of money are swindled for alleged flight tickets or to pay fees. Skilled crooks with suitable victims can 'earn' astonishing sums of money,” explains social media consultant Felix Beilharz to pressetext.

Blackmail is also an option that occurs again and again. “After a phase of getting to know each other and building up trust, you arrange to meet up for video sex, for example. The scammer records the whole thing and threatens to publish the video if no amount of money is paid. I am personally aware of such a case that completely ended the career of the person concerned,” warns Beilharz urgently. The platforms are of course trying to take countermeasures. According to the expert, artificial intelligence is a promising approach.

Certain patterns recognizable

The research team trained a neural network to analyze dating profiles. The data was used from the website Dating 'N More . This consists of 14,720 real profiles and 5,402 fake pages. The scientists used 60 percent of the data for training, 20 percent for validation and 20 percent for testing purposes. Using the fake profiles, the researchers were able to identify certain patterns.

The system consists of a neural network for analyzing profile pictures and a speech-generating neural network for examining the text in people's biographies. The focus is on a person's age, gender, ethnicity, occupation, marital status and place of residence. Of the fake profiles, around 60 percent were men and 40 percent were women. These fake male profiles had an average age of 50 years, while their female counterparts had an average age of 30 years.

There are common patterns that scammers use to make their profiles more attractive, regardless of what gender they claim to be. Mock profiles contain more images and text with “emotional language,” such as people describing themselves as “caring,” “passionate,” and “loving.”