‘Is my customer human?’ Banks must rely on AI to uncover automated fraud

By Xin Ren, Senior Director of Data Science, Feedzai

Generative AI (GenAI) is changing the game for businesses and becoming a new tool for fraudsters. Regulating AI and expanding the role of technology actors in the financial services sector is essential to curb the evolution of fraud.

According to the 2024 State of Scams Report by GASA & Feedzai, banking fraud and scams led to losses of US$1 Trillion Globally. These significant losses highlight that it is more important than ever for banks to stay one step ahead of criminals. Banks need to shift from responding to fraud to preventing it.

Generative AI paves the way for automated fraud and financial crime

With existing tactics like social engineering, access to sensitive information released on the dark web due to data breaches, social media, and stolen banking data, fraudsters can massively enhance their scams.

Not only can fraudsters create new identities faster, but GenAI also lends these identities greater credibility. The ability to create fake images, videos, or even false voice recordings allows fraudsters to construct a character with its own fabricated identity from scratch.

With the capacity to pass authentication checks, generative AI becomes a powerful tool for financial crime.

In this context, banks will be tempted to question the authenticity of every interaction as these tools become more widespread and advanced.

Fraud as a service is gaining ground

Thanks to generative AI, call centers can now quickly gather information about their targets, learn organizational operations, and tailor attacks to specific banks. This is particularly concerning for new account fraud and account opening requests.

Criminals can use GenAI tools to learn the different layouts and steps of a bank’s screens. With this knowledge of how different organizations function, criminals can write scripts to quickly fill out forms and create seemingly credible identities to commit account opening fraud. Banks will no longer only need to ask, “Is this the right person?” but also, “Is my client human or AI?”

Embracing the power of AI is essential

Businesses need advanced fraud prevention tools to protect themselves against AI-driven threats. AI-powered fraud detection systems enable organizations to analyze vast amounts of transaction data in real-time, uncover hidden patterns and warning signals that traditional methods might miss. Alerts generated by AI should include clear explanations so that human analysts can understand the reasoning behind potential issues and make informed decisions.

AI algorithms can be biased and require constant monitoring and improvement. This is why human expertise remains a crucial element in AI-based decision-making.

Banks can use AI to take proactive measures by predicting future risks through risk assessment. However, banks using AI for credit evaluation or fraud detection, among other applications, must ensure their systems are effective, ethical, transparent, and accountable.

Expect more FinServe collaboration

Collaboration is becoming key, as banks are increasingly joining forces with other financial institutions, including fintechs and regtechs. The goal is to share data and knowledge to strengthen defenses against cross-border fraud systems. However, banks are hesitant to share information if they fear exposing themselves to legal issues.

To improve data sharing and collaboration, regulators must clarify or ease their stance toward banks. The goal is collaboration, but greater clarity is needed at the highest levels to ensure data is shared correctly.

Financial institutions can also optimize resource allocation with data-driven insights, focusing their attention on high-risk cases. This reduces the need for exhaustive manual investigations into every transaction, allowing teams to prioritize their efforts effectively. As a result, organizations can enhance efficiency and reduce costs by preventing the most significant fraud cases.

AI and machine learning: A shield against fraud

Banks need AI and machine learning to detect and prevent fraud in real-time. Fraud analytics not only help reduce potential losses but also build customer trust in their banks.

Fraud analytics combines artificial intelligence (AI), machine learning, and predictive analytics for advanced data analysis, enabling banks to process large volumes of data and quickly derive insights to respond in real-time to suspected fraud.

Based on our experience with US banks, these banks are now able to detect half of potentially fraudulent transactions. However, the use of AI and machine learning has enabled them to detect 60% of fraud, preventing millions in potential fraud losses.

Additionally, GenAI capabilities have reduced false positives by 40 percent, allowing banks to provide a more transparent and frictionless customer experience.

Conclusion

In the era of big data, banks can no longer rely solely on traditional rule-based systems to detect fraudulent transactions. Fraudsters quickly learn a bank’s rules and find ways to commit fraud without being detected. Each new fraud tactic brings new learning, pushing banks into an endless game of cat and mouse.

For banks and financial institutions, the evolving regulatory landscape around AI presents both challenges and opportunities. On one hand, institutions must be agile in updating their AI-driven processes to comply with new guidelines while considering potential liabilities. On the other hand, adherence to these principles can strengthen customer and stakeholder trust, which is highly valuable in the financial world.

About the author

Xin Ren is Senior Director of Data Science at Feedzai. She has been working in the Financial industry for over 10 years and specialized in delivering AI-based strategy and bringing data science best practices to the clients.

Article Topics

biometrics  |  digital identity  |  Feedzai  |  financial crime  |  financial services  |  fraud prevention  |  generative AI