Industry

Erste Bank develops AI-driven fraud detection as online scams rise

Erste Bank reported over 7,000 attempted frauds in the first quarter of the year, with attackers targeting 2.8 billion forints; the bank’s systems detected eight out of ten attempts and blocked about 2.3 billion forints in potential losses.

Erste Bank said that in the first three months of the year its staff detected more than 7,000 attempted frauds. Attackers sought to extract a total of 2.8 billion forints, while the bank’s security systems identified about eight out of ten attempts and thereby prevented roughly 2.3 billion forints from being lost.

Methods and financial impact

Statistics show a clear difference between card fraud and transfer-based scams. Criminals most often aimed to obtain card details: in about 6,000 cases they targeted card data, representing roughly 0.5 billion forints in potential loss. Transfer-based frauds were less frequent but more damaging per incident: the average loss per transfer-case exceeded 1.6 million forints at the start of the year, while the average loss for a successful card incident was around 84,000 forints.

Zsiga Krisztina, Deputy CEO for Risk Management at Erste Bank, highlighted that the highest risk occurs when criminals gain direct access to a customer’s online banking account.

Psychological manipulation and disguised schemes

The most effective attacks rely on psychological manipulation. Perpetrators often call victims and impersonate bank staff, security experts, or employees of the Magyar Nemzeti Bank. Recently, fake investment offers have caused major harm: scammers initially persuade victims to invest small amounts, then use professionally designed fake websites to display non-existent returns and encourage further transfers. Some criminals even train victims what to say to a real bank representative if an alert call is made.

According to Magyar Nemzeti Bank data, phishing remains the leading attack vector: criminals send mass fake delivery-service or authority links, online marketplace scams trap many users, and numerous people install remote-access programs on their devices at the attackers’ instruction.

Bank protections: app controls, limits and artificial intelligence

Domestic banks are continuously strengthening their defenses. Besides a real-time fraud-monitoring system, Erste is developing a self-learning network based on artificial intelligence aimed at better identifying fraud patterns and blocking fraudulent transactions.

The bank’s mobile application also includes multiple protective features: account holders can set transaction limits themselves and receive instant notifications of any data or setting changes. Banks additionally run information campaigns, podcasts and targeted messages to raise public awareness.

National picture and responsibility for losses

The Magyar Nemzeti Bank’s report for the fourth quarter of 2025 shows that cybercriminals extracted 5.6 billion forints nationwide over a three-month period. The attack surface is large: by the end of 2025 some 87.7 percent of domestic retail bank accounts were accessible online.

A painful issue is responsibility for losses. Central bank figures indicate that in successful scams the lost funds are overwhelmingly borne by customers: 68.3 percent of card-fraud losses and 89.8 percent of transfer-related losses were charged to individuals in the referenced period. Banks are often unable to provide reimbursement when customers themselves disclose secret codes, authorize transactions, or willingly install spy software on their devices.

The trends point to the need for continued investment in fraud defenses — including artificial intelligence, user notifications and ongoing education — to reduce the risk and scale of losses suffered by retail customers.