Within the complex modern financial system, banks are the guards that protect the assets and transactions of many people. Yet it is because of this status that they are the perfect victims of competent criminals. The banking sector is no stranger to financial scamming, and according to Kathy Stokes, director of fraud prevention programs at AARP, financial fraud is at a crisis level.1 With advanced technology, thieves can capture customer information that can be used to impersonate an individual and access their bank accounts. There were nearly 1.7 million reports of identity theft in 2021 alone.2
Given the increasingly complicated and frequent scams that challenge the security and integrity of banks worldwide, they must integrate these tools into their fraud prevention systems, and quickly.Contending with bank fraud is a constant challenge. Banks experienced a 41% increase in fraud incidents between 2021 and 2022,3 and Automated Clearing House (ACH) fraud increased by 5%.4 Meanwhile, fraud attacks via mobile wallets quadrupled.5 These types of fraud cost US consumers more than US$7 billion in the first three quarters of 2023, a 5% increase over the same period in 2022, according to the US Federal Trade Commission.6
The arsenal to combat such fraud exists within the very technology that emboldens it: digital transformation and artificial intelligence (AI). Given the increasingly complicated and frequent scams that challenge the security and integrity of banks worldwide, they must integrate these tools into their fraud prevention systems, and quickly.
The digitalization of banking has led to several outcomes, both positive and negative. While technology has improved the relationship between customers and banks and made transactions seamless and more efficient, it has also created fertile ground for fraud. The Association of Financial Professionals (AFP) and J.P. Morgan posit that checks and wire payments are the most common targets for fraud.7 The increase in check fraud is attributable to the use of mobile check deposits, which spiked 41% between 2020 and 2021.8 However, the development of new financial technologies leads to the launch of new and more advanced types of fraud schemes. Account takeover, advance fee fraud, identity theft, check fraud, ACH fraud, first-party fraud, and wire fraud are some of the most common attacks against banks and their customers.
The spike in fraud incidents and the amounts of money involved have forced banks to rethink and restructure the way they prevent financial crime. Machine learning (ML), digital transformation, and AI have become the main constituents of financial fraud prevention. Through the adoption of AI, financial institutions can quickly screen large data sets, discover patterns of potential fraudulent activity, and institute security mechanisms that can adapt and react to the fast-changing threats. For example, Citibank deployed AI to detect account takeover.9 The AI system is designed to detect unusual payment behavior or account usage patterns and alert the proper contact, thereby mitigating and/or reducing any financial loss.10
The Role of Digital Technologies in Banking Transformation
The influx of digital technology has had a massive impact on financial markets. Applying digital technologies such as cloud computing, data analytics, and mobile technologies and programs can make banks’ workflows more efficient and present more opportunities to interact with clients. These technologies have fueled the growth of electronic banking. This revolution necessitates the rebuilding of banks’ security procedures through a constant and comprehensive reassessment of threats from hackers and identity thieves.
Fraud has grown in parallel to the expansion of digital banking channels. Financial institutions can improve their security architecture through the application of several algorithms and encryption technologies:
- Digital identity verification systems, which are biometric-based and use multifactor authentication (MFA), have played a significant role in preventing identity fraud and illegitimate access to financial information.
- The fraud department of a credit union leveraged Hitachi’s AI bot to evaluate suspicious mobile deposits with the objective of preventing fraud.11 The AI bot prevented fraud by analyzing mobile deposits below the threshold set by clients, which resulted in increased loss prevention amounts and customer confidence.
- The introduction of blockchain has created an immutable record of transactions, which greatly reduces fraud because it provides transparency and traceability.
AI and ML (both supervised and unsupervised) algorithms act as gatekeepers to halt fraudulent activities in real time. These systems investigate sequences and identify variations that do not coincide with routine customer behavior, allowing banks to react expeditiously in cases of fraud. According to research by PYMNTS, 88% of bank executives say that reducing fraud is critical to maintaining merchant processing revenue, and most are using ML and AI to do so.12 Nearly all the banks surveyed (98%) use AI, and 60% use it as their primary weapon.13 Additionally, 27% of bank executives say that rules-based algorithms are the most important tool for combatting various types of bank fraud.14
AI’s Contribution to Financial Fraud Detection
In the realm of financial security, AI is at the forefront, providing algorithms to help handle transactional data and separate criminal financial activity from legal transactions. This is accomplished largely through ML and predictive analytics—two branches of AI that process and analyze high volumes of data to find patterns indicating fraud. After an ML algorithm is trained on historical transaction records, it becomes highly proficient in recognizing the normal pattern of each customer's behavior as well as fraudulent imitations of it, an ability that increases with each transaction investigated. Next, predictive analytics, based on statistical algorithms, models potential fraud based on historical trends. It analyzes factors such as transaction sequences and exploratory user conduct to predict and prevent future fraud.
AI makes fraud detection faster, more reliable, and more efficient, succeeding where traditional fraud detection models have failed. AI is very effective in fighting identity theft, document forgery (e.g., fake IDs, forged signatures, fake credit card and loan applications), credit card theft, and phishing attacks by gathering, processing, and categorizing historical data.
Consider several examples:
- When Danske Bank, the biggest bank in Denmark, experienced a major money laundering crisis, it responded by implementing a tool built by AI and made by Teradata.15 The system uses ML algorithms that enable it to monitor and analyze customers’ transactions in real time. As a result, false positives decreased and complex fraud schemes were discovered where previous systems had failed.
- Another exceptional case of the use of AI in banking is J.P. Morgan Chase’s Contract Intelligence (COIN) program. It analyzes commercial loan agreements by itself—a job that used to consume 360,000 lawyer-hours and loan officer–hours annually.16 COIN can eliminate loan servicing mistakes caused by human error in the conventional language of loan documents by interpreting the content and meaning with the help of ML.
- At Wells Fargo, AI has been used to detect check fraud, greatly improving the bank's security.17 AI found inconsistencies in checks presented at different locations. Using established patterns, the system recognized suspect handwriting and signatures on checks that were part of a hoax. AI’s early detection of these fraudulent checks prevented a large financial loss.
- At HSBC, AI detected and prevented fraud by monitoring anomalous transaction patterns.18 The bank’s technology found suspicious activity in a customer's account, including large transfers to an offshore account. These transactions were tagged for AI rating and the investigation revealed a compromised account. The AI alert prompted HSBC to freeze the account, stopping any further fraudulent activity. This case shows how AI can protect against complicated financial schemes such as identity theft and unauthorized access to accounts.
In addition to these examples, the use of infographics can also provide a compelling demonstration of AI’s role in the detection and prevention of fraud. Infographics can pictorially represent the AI workflow: data intake (data ingestion), data processing, detection of anomalies, notification of anomalies (issuing an alert), and finally human verification and responsibility (human intervention). This type of illustration conveys the complex nature of intellectual exploitation in banking, thus facilitating understanding. Visualizations improve the ability to identify relationships, structures, and patterns of suspicious transactions in large quantities of complex data, enabling fraud detection and prevention.19
Overcoming the Challenges of Using AI to Fight Bank Fraud
The introduction of AI in banking is accompanied by its own set of challenges. At the top of the list is the high cost of installing such technologies, which may be a barrier to small enterprises. In addition to the financial cost, it takes time and resources to train AI models and the personnel needed to manage and analyze AI systems efficiently.
Other challenges may involve the process of integrating AI into legacy systems that are already in place. Some banks still use operating systems that are not integrated with technologies that support AI. This technical barrier cannot be easily overcome. The rigidity of these systems can impair an enterprise’s ability to process the real-time data necessary for AI activities.
Data privacy is another issue. The financial industry is an extremely well-regulated area, and any AI solution must adhere to a variety of data security laws such as the EU General Data Protection Regulation (GDPR) and the US Data Privacy and Protection Act. AI's learning process involves collecting large amounts of data, which may raise concerns about consumer privacy and data safety.
These challenges can be addressed in a multidisciplinary manner. The financial barrier can be overcome by forming alliances with fintech businesses and applying as-a-service systems to spread out the cost over time. Financial institutions must also ensure that employees with the requisite technical skills work alongside AI to guarantee that the human workforce is not displaced by technology, but rather supported by it.
To address the issue of system integration, a layer-by-layer approach is necessary. The compatibility of the two systems must be handled first, followed by the gradual implementation of AI. Adaptation of banking norms can be viewed as the gateway through which new technologies acquire better connectivity with old banking systems.
To encourage data privacy, it is important to use reliable encryption and ensure that AI use protects sensitive data and aligns with the developer’s original intent. Tactics such as federated learning can be effective, in which case data is studied at the same time across all user devices so that it does not leave the device. To implement AI for fraud prevention, banks must commit to ensuring data privacy and security. This is critical to prevent data breaches and protect customers' confidential information.
Conclusion
AI and digital transformation play a crucial role in reducing workload, enhancing efficiency, improving customer service, and strengthening cybersecurity in the financial sector. Today’s AI powerhouses, especially ML and predictive analytics, have become important players in unmasking and thwarting illegal activities and improving financial security.
Looking ahead, the banking industry is expected to adopt AI at warp speed. Superior computing facilities such as cognitive computing, deep learning, and natural language processing are in sight. These innovations are expected to generate faster and more efficient detection systems that emphasize the minute details of analyzing and predicting fraud with precision.
Although the introduction of AI in the banking industry presents some challenges, including high costs, technology incompatibility, and privacy issues, solutions are available. Strategic partnerships, a phased approach to technology adoption, a proper cybersecurity scheme, and alignment with applicable regulations are some of the measures that can be implemented to navigate this complicated space.
Endnotes
1 AARP, “AARP Report: Americans Agree that Fraud Is at a Crisis Level,” 17 May 2023, https://press.aarp.org/2023-5-17-AARP-Report-Americans-Agree-Fraud-is-at-Crisis-Level
2 Caporal, J.; “Identity Theft and Credit Card Fraud Statistics for 2024,” The Ascent, 29 February 2024, https://www.fool.com/the-ascent/research/identity-theft-credit-card-fraud-statistics/
3 PYMNTS, Money Mobility Tracker, July 2022, https://www.pymnts.com/tracker/money-mobility-banking-fraud-prevention-aml/
4 Association for Financial Professionals, 2022 AFP Payments Fraud and Control Report Key Highlights, 2022, https://www.jpmorgan.com/content/dam/jpm/commercial-banking/insights/cybersecurity/highlights-afp-2022-payments-fraud-and-control-report.pdf
5 DataVisor, “12 Most Common Types of Bank Frauds,” https://www.datavisor.com/wiki/types-of-bank-frauds
6 DFederal Trade Commission, “As Nationwide Fraud Losses Top $10 Billion in 2023, FTC Steps Up Efforts to Protect the Public,” USA, 9 February 2024, https://www.ftc.gov/news-events/news/press-releases/2024/02/nationwide-fraud-losses-top-10-billion-2023-ftc-steps-efforts-protect-public
7 Association for Financial Professionals, “2024 AFP Payments Fraud and Control Survey Report,” 2024, https://www.afponline.org/publications-data-tools/reports/survey-research-economic-data/Details/payments-fraud/
8 Op cit DataVisor
9 Crosman, P.; “Citi Deploys AI to Detect Real-Time Fraud, Errors in Payments,” American Banker, 21 December 2018, https://www.americanbanker.com/news/citi-deploys-ai-to-detect-real-time-fraud-errors-in-payments
10 Boon, S.; “How AI and ML Are Used in Payment Fraud Detection (16 Use Cases),” Nomentia, 3 May 2024, https://www.nomentia.com/blog/ai-machine-learning-in-fraud-detection
11 Wingard, L.; “Fraud Management in Banking: Detection, Prevention and More,” Hitachi Solutions, https://global.hitachi-solutions.com/blog/fraud-prevention-in-banks/
12 Op cit DataVisor
13 Ibid.
14 Ibid.
15 Teradata, “Danske Bank and Teradata Implement Artificial Intelligence (AI) Engine that Monitors Fraud in Real Time,” 23 October 2017, https://www.teradata.com/press-releases/2017/danske-bank-and-teradata-implement-ai
16 J.P. Morgan, “About Coin Systems,” https://www.jpmorgan.com/onyx/coin-system
17 FICO, “Wells Fargo Enhances Fraud Protection and Customer Experience Using FICO Solutions,” 16 March 2022, https://www.fico.com/en/newsroom/wells-fargo-enhances-fraud-protection-and-customer-experience-using-fico-solutions
18 Chaturvedi, P.; “Financial Services Trends Driving Innovation,” HSBC, 4 February 2024, https://www.business.us.hsbc.com/en/insights/growing-my-business/financial-services-trends-driving-innovation
19 yWorks, “Fraud Detection Through Visualization,” https://www.yworks.com/pages/fraud-detection-through-visualization
Maduabuchi Christopher Okonkwo| BSC, MSC, MBA, Certified Scrum Master
Is an IT manager with more than 16 years of experience in the financial services, e-commerce, and health sectors specializing in digital transformation, cybersecurity, and artificial intelligence (AI). He can be reached at www.linkedin.com/in/christopher-okonkwo-550a8787.
