AI agents: The biggest weakness in the fight against financial fraud

Whether people trust digital financial services affects economic stability, social cohesion and the security of everyday life. And this trust will be under enormous pressure in 2026. Fraud and financial crime have long since developed into an interwoven system that systematically overwhelms traditional defense mechanisms. According to the Global Financial Crime Report 2026 by Nasdaq, an estimated USD 4.4 trillion in illicit funds flowed through the global financial system in 2025 alone - an increase of 42 percent compared to 2023. Fraud accounted for an estimated USD 579.4 billion; the number of unreported cases is likely to be considerably higher.

These figures form the background to a broad-based study by Biocatch, a company that combats fraud and financial crime by analyzing patterns in human behavior. The company commissioned Team Lewis Research to conduct an online survey of 1,440 fraud management, financial crime prevention, risk and compliance professionals from 25 countries across North America, Europe, the Middle East and Africa, Latin America and Asia Pacific.

Rising losses despite high investments

The results are alarming: 81% of respondents report an increase in fraud attempts against their customers - compared to 71% in the previous year. The trend in losses is even clearer: the proportion of those reporting increasing fraud losses year-on-year has risen from 59 to 76 percent. Almost half of the institutions surveyed state that they spend more than 10 million US dollars annually on combating fraud and financial crime. Nevertheless, more than three quarters report rising losses - in most cases, existing resources are not sufficient to keep pace with the growing threat.

The most common reasons for the increase: 59% of respondents cite scam attempts via real-time payments, 45% the increasing use of AI and automation by criminals and more cases of bank transfer fraud where victims pay themselves. Criminal networks act like professionally structured companies - and are not bound by regulations. 59 percent of respondents say that criminal organizations are adapting faster than financial institutions.

AI: tool and weapon at the same time

The rapid spread of AI tools has significantly exacerbated the threat situation. Scams are becoming hyper-personalized, more convincing and barely recognizable for those affected. 88% of respondents are convinced that AI has already made scams much more sophisticated and dangerous. Criminals are using generative AI to automate and scale social engineering attacks on a massive scale - with deceptively real phishing emails, deepfake audio and video content and fake documents that outsmart both traditional detection systems and human verification.

Fraud management experts most frequently observe: deepfake-supported social engineering and identity deception (50 percent), automated phishing and spear phishing attacks (48 percent) as well as automated money laundering and transaction fraud (44 percent). But AI is not just a weapon - respondents also see considerable potential for fraud prevention: they cite real-time transaction monitoring and anomaly detection (40 percent), behavioral analysis and pattern recognition (34 percent) and identity verification and authentication (27 percent) as the greatest benefits.

AI agents as the biggest weak point

84% of respondents see AI agents as the biggest gateway for fraudsters in the coming year. And the threat is already a reality: 80 percent of respondents state that their institution has already experienced attacks in which AI agents were used. Financial institutions in Latin America (89 percent) are particularly affected, followed by Europe, the Middle East and Africa, the Asia-Pacific region (79 percent each) and North America (75 percent).

The growing use of AI agents in banking creates a new challenge: in future, banks will have to be able to distinguish between three scenarios - a genuine customer carrying out legitimate banking transactions; a fraudster who has taken over someone else's account; and a genuine customer acting under the manipulation of a scammer. 72 percent of respondents expect that it will be very difficult to recognize whether an AI-supported transaction is legitimate or controlled by criminals. At the same time, around four out of five respondents (82 percent) say they are already prepared to assess the intentions behind a transaction - even if it is carried out by an AI agent.

Information exchange as the key

Fraud, scams and money laundering are now an interconnected, systemic problem. Financial institutions worldwide are increasingly recognizing that combating this systemic threat requires a systemic solution. 85 percent of respondents are convinced that real-time information sharing between banks would significantly improve fraud detection. 86 percent say that real-time information about the destination account of a transfer would help to prevent scams in good time.

However, implementation fails in practice due to various hurdles: Data protection regulations (26 percent), fear of legal liability (19 percent) and concerns about confidential information (19 percent). Nevertheless, according to the survey, almost three quarters (74%) of the institutions surveyed exchange information on suspicious accounts with other financial institutions at least once a week.

Trust as a currency

The consequences of fraud go far beyond financial losses. 74 percent of the bank managers surveyed often observe negative effects on trust or customer loyalty after a fraud or scam incident. 68 percent are convinced that their institution's handling of fraud prevention and refunds has led to a loss of customers. While more than half (56 percent) attribute this migration to a lack of refunds, the remaining 44 percent see the cause in too much friction and unnecessary hurdles in the customer process.

Gadi Mazor, CEO of Biocatch, comments on the development: «AI is beginning to fundamentally change how customers interact with e-commerce offerings and financial institutions - and it will also change how criminals commit fraud and other financial crimes. As digital interactions become faster, more automated and increasingly agent-driven, we need to move beyond static identity verification to a deeper, more immediate understanding of behavior, intent and trust.»

Financial institutions that keep an eye on their reputation in real time, invest in innovative technologies, share information with other institutions and communicate openly with their customers are better positioned to build and maintain customer trust in the long term. To survive in the increasingly digital race against criminals, real-time information sharing and closer cooperation - across institutions and national borders - are urgently needed.

Source: www.biocatch.com

This article originally appeared on m-q.ch - https://www.m-q.ch/de/ki-agenten-die-groesste-schwachstelle-im-kampf-gegen-finanzbetrug/