A new artificial intelligence system can now help insurers and the police identify fake robbery claims.

Developed by researchers from Cardiff University and the Charles III University of Madrid, the AI system “VeriPol” utilizes automatic text analysis and machine learning to identify false statements. The developers claim that VeriPol is quite accurate with its verification process, “with over 80% accuracy.”

The AI tool is capable of analyzing written statements to recognize patterns which are typically associated with false statements. Statements the AI looks out for include the kinds of items allegedly stolen, the descriptions given of the purported attackers, and even the “finer points” of an incident, the researchers explained.

The researchers added that the common themes that indicate that a robbery claim is fake include a focus on the items stolen rather than the incident itself, lack of detail from the reporter suggesting fabrication, limited details about the attacker, lack of witnesses, and a failure to contact either law enforcement or a medical professional immediately after the incident.

Gadgets and other similar high-end items such as iPhones and Samsung phones were often associated with fake reports, while jewelry and bicycles are usually the items reported stolen in real robbery cases.

Several law enforcement agencies in Spain are now using VeriPol, ZDNet reported. The Spanish National Police even conducted a trial of the AI tool, letting it analyze more than 1,000 police reports.

Related stories:
Evolution in insurtech continues, backed by funding from industry giants
Why understanding AI is “non-negotiable” for insurers