AI as a detective: How neural networks help insurers fight fraud

1045
Max Abazov Correspondent

Over the past year, Kazakhstan has seen a sharp surge in cyber fraud. According to research company URTG (United Research Technologies Group), nearly every second Kazakh citizen has received suspicious phone calls from someone pretending to be a bank or a police officer. Unfortunately, the number of victims is also on the rise.

Data from the National Bank of Kazakhstan shows that in the first eight months of this year alone, Kazakh citizens transferred 16.8 billion tenge to fraudsters - almost 30% more than in the same period in 2024. Between January and August, more than 92,000 cases of cyber fraud were reported, with 8,400 confirmed as proven scams. Banks managed to block 2.6 billion tenge worth of attempted theft and refund 339 million to customers, while police identified 14,761 suspects.

© Sputnik

As technology advances, so do the methods used by scammers. URTG’s report notes that around 51–55% of Kazakh citizens can recognize typical fraud schemes such as phishing, smishing, and vishing. Awareness among corporate professionals is much higher - about 70–85%. However, public understanding of more complex and technology-based threats such as deepfakes remains limited: only 28% among the general population and up to 60% in the corporate sector.

Fraudsters do not shy away from targeting large companies either. In the insurance industry, there are frequent cases when clients or their representatives submit claims for incidents that never actually occurred. To counter such schemes, insurers are implementing specialized anti-fraud systems that automatically check claims for inconsistencies and take appropriate measures when fraud is suspected.

Azamat Kerimbayev / Photo: © Valeriy Ayapov

According to Azamat Kerimbayev, Chairman of the Board at Freedom Insurance, the company relies on Sergek system data to detect fraudulent claims:

“We use data from the Sergek system to verify whether an incident really took place. For example, the system can check footage from road cameras to confirm or disprove a reported accident. This helps uncover staged collisions - when a damaged car is brought to the scene and presented as if the crash just happened,” he explains.

Kerimbayev adds that Sergek data also helps insurers make better-informed decisions when assessing premiums and payouts:

“The system analyzes how frequently and in which areas a vehicle has been used under a specific license plate. This information doesn’t serve as grounds for denying a claim, but it helps form a complete picture of the vehicle’s history. If discrepancies appear, we invite the client for a meeting to clarify the details. When necessary, we can also request verification from law enforcement agencies.”

press service of DP Almaty

At Freedom Insurance, the same technology is used when issuing voluntary auto insurance policies:

“Sergek analyzes how often the vehicle is driven, on which days, and along what routes. This data is part of a scoring process based on AI algorithms, which also evaluates potential fraud risks. For instance, if a car hasn’t been used for a long time and suddenly appears on the road, and the owner applies for a CASCO policy the same day - that raises questions. Combined with Sergek data, this analysis helps us build a more accurate risk profile, offer fair pricing, and protect both the company and honest clients from fraud,” the expert concludes.

By integrating Sergek’s intelligent monitoring system and AI-powered analytics, the insurance sector in Kazakhstan is reaching a new level of transparency and trust. These technologies not only help detect and prevent fraud but also ensure that customers receive fair and data-driven insurance evaluations, strengthening confidence between insurers and policyholders.

DKNews International News Agency is registered with the Ministry of Culture and Information of the Republic of Kazakhstan. Registration certificate No. 10484-AA issued on January 20, 2010.

Theme
Autoreload
МИА «DKnews.kz» © 2006 -