AI-driven payment fraud causes major financial losses in Europe
European payment fraud reached 1.2 billion euros in 2024, with AI-powered attacks including deepfake voice imitation and IBAN forgery increasing sophistication. While financial institutions deploy AI systems for fraud detection with up to 95% accuracy, criminals exploit the same generative AI technology faster than regulators can adapt.
What happened
In 2024, payment fraud across Europe reached approximately 1.2 billion euros, with 382 million euros linked to manipulations. Fraudsters now use generative AI to create false payment orders, imitate executive voices through deepfake audio, forge IBANs, and conduct targeted phishing attacks.
Nearly nine out of ten financial institutions use AI models to detect transaction anomalies and assess risk, achieving detection rates up to 95%. Simultaneously, criminals employ the same generative AI technology for fraud, adapting faster than regulatory measures can address.
Reported impact
Financial institutions report treatment costs have been halved through AI deployment. A survey by Boston Consulting Group found that only one out of four banks feels ready to safely integrate generative AI models into their own protection systems.
Index analysis
Federated learning, or distributed machine learning, allows banks, fintech companies and enterprises to train AI models collaboratively without sharing sensitive data. This approach increases fraud detection precision by 20% on average and complies with GDPR by sharing only updated parameters rather than raw data.
Governance controls
As of October 9, 2025, the European Union introduced new rules on instant transfers requiring mandatory beneficiary verification (Verification of Payee, VoP), an automated check matching name to IBAN to reduce errors and fraud. Experts note that fraud evolves faster than regulations, requiring collaboration between enterprises, banks and public authorities within a framework of transparency and oversight.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.