Financial institutions lose billions every year to fraud. Singapore’s OCBC Bank reported that $13.7 million was lost in a wave of phishing scams last year. Since then, the Monetary Authority of Singapore (MAS) has established new measures to enhance digital banking security. In addition, insurance fraud – including health insurance claims and motor insurance scams – has more than tripled, according to the Minister for Law and Home Affairs K. Shanmugam, underlining weak fraud detection mechanisms in the region.
Many organisations still rely on legacy fraud detection systems that aren’t adequate against large-scale phishing campaigns to dupe unsuspecting consumers into sharing personal information. Left undetected, these scams leave a trail of destruction, impacting organisational reputation and damaging public trust in institutions.
The Singapore Police Force released data showing that scam victims in Singapore lost over $660 million in 2022. The latest Singapore Cyber Landscape published by the Cyber Security Agency of Singapore also revealed that phishing attacks continue to pose an increased threat to organisations and individuals. The report cited that the Singapore Cyber Emergency Response Team received around 8,500 phishing attempts in 2022, more than double the 3,100 cases handled in 2021. Over 80% of reported sites pretended to be legitimate banking and financial services entities.
Given these alarming statistics, fraud awareness and detection are critical for businesses of all sizes.
Traditional fraud detection methods often fail to minimise losses since they perform discrete analyses susceptible to false positives and negatives. As a result, sophisticated fraudsters have developed various ways to exploit the weaknesses of discrete analysis.
Many companies are turning to artificial intelligence to identify and prevent fraud, using machine learning to identify suspicious behaviour. While useful in detecting the most likely cases, it isn’t effective when sifting through billions of data points.
See also: Keys to achieving human-centred automation testing
Uncovering fraud patterns in real time
Graph technology has emerged as a critical enabler in fraud detection and prevention. Unlike conventional fraud detection models, it allows organisations to identify and predict fraud and anomalies at scale using connections between data points. This provides better context and the ability to link fraudulent events and activities – enabling organisations to have better insights and fraud detection capabilities.
Furthermore, algorithms and machine learning techniques can identify fraudsters based on similarities or reveal fraud rings or money launderers. “Guilty by association” scores are generated based on the quality, quantity, and distance of someone’s relationship with suspicious entities. Once the pattern of a fraud ring is generated, a similar algorithm can use the pattern to detect other potential rings and their participants.
See also: Human element still important for effective mass communication
Leading fintech company Banking Circle used traditional rule-based strategies – such as assigning risk scores by searching for certain words, accounts, and locations – to detect fraud and money laundering. These alerts were then sent to fraud analysts for manual review, making the entire process slow, cumbersome and expensive.
With graph and machine learning techniques, Banking Circle was able to scale its operations effectively with remarkable results. The company processed 155 billion euros ($228bn) payments in 2020. It approached 100 million annual bank transfers by the end of 2021, with a 25% drop in false negatives and a massive reduction in overall alerts escalated for manual review.
A powerful enabler
Graph technology can equip Singaporean organisations to identify patterns and create a feedback loop of insights, resulting in a direct return on investment. The best part is its highly effective fraud detection capabilities that evade traditional approaches. It also has higher accuracy in escalating cases for manual review through scalable, flexible solutions.
As business processes become faster and more automated, the time to detect fraud is narrowing, amplifying the need for real-time solutions. It’s clear that traditional technologies are not designed to detect elaborate fraud rings. This is where graph databases can add value by analysing connected data points and uncovering their hidden relationships.
Graph technology was instrumental in the Panama Papers investigation by the International Consortium of Investigative Journalists (ICIJ), where 40 years’ worth of confidential documents were reviewed to unmask fraudulent entities. Graph technology was invaluable in exploring networks and finding connections that helped journalists dig deeper into the biggest data leak of all time. As a result, governments globally have recouped over US$1.36 billion ($1.75bn) in back taxes and penalties.
In 2021, ICIJ worked on the largest investigation in journalism history, the Pandora Papers. The investigation leveraged graph technology to expose tax and secrecy havens and money laundering activities of over 330 politicians, including 35 current and former world leaders, 130 Forbes billionaires, as well as celebrities, fraudsters, drug dealers, royal family members and leaders of religious groups across the globe. With almost 12 million records from 14 different offshore services firms, graph technology enabled ICIJ to identify hidden connections and expose the truth about wealthy elites from more than 200 countries and territories.
With scammers showing no signs of relenting, organisations will need real-time fraud detection capabilities to make faster, more accurate decisions to mitigate losses. Graph technology’s preventative approach to cyber security is a powerful tool to help combat cybercrime in the region as it becomes harder to detect.
Nik Vora is the vice president of Apac at Neo4j
