Sigma Synthetic Fraud v4 uncovers multiplex synthetic-specific features
Socure has launched Sigma Synthetic Fraud v4. The product uses advanced machine learning and diverse, third-party and network feedback data to uncover patterns linked to insidious synthetic identity fraud.
The Deloitte Center for Financial Services expects synthetic identity fraud to generate at least $23 billion in losses by 2030.
Synthetic identity fraud is a financial crime where a real person’s information is stolen and combined with other falsified personal information to create a fictitious identity, further used for fraudulent purposes. After a perpetrator opens an account using the synthetic identity, they typically build up a positive credit score, open multiple accounts, and often appear to be good customers while going undetected until they decide to cash in, or “bust out” by using up all available credit lines and disappearing.
Socure accurately detects and stops synthetic fraud at onboarding before the fraudster can act nefariously in the financial ecosystem. According to a comprehensive study, Socure estimates that synthetics make up 1-3% of open accounts at U.S. financial institutions.
Sigma Synthetic Fraud v4 draws from diverse “Proof of Life” data sources including property records, driver’s licenses, and educational data adding a new dimension of accuracy so organizations can confidently verify younger and immigrant demographics with a limited digital footprint. Without these types of proof of life data sources, these segments of the population may otherwise appear to be synthetic fraudsters and be shut out of the financial ecosystem.
“Synthetic fraud cannot be accurately detected with rules-based systems or third-party fraud solutions,” said Yigit Yildirim, SVP, Fraud and Risk Products at Socure. “Socure’s AI engine analyzes anomalies to uncover multiplex synthetic-specific features that distinguish legitimate thin-file consumers from synthetic fraudsters with high accuracy in real-time — and without causing friction for good users.”
Synthetic identity fraud occurs when criminals blend genuine and falsified information to create new, fictitious identities to fraudulently apply for loans, credit, government benefits, or move illicit funds. As fraudsters’ AI-supported schemes become more sophisticated, differentiating malicious synthetic behavior from that of good consumers is more tangled than ever and has made it the fastest-growing form of financial crime in the United States.
Per incident, synthetic fraud can cost 10 times more than third-party identity fraud. The “profit” per synthetic fraud opportunity is much higher, such as with benefits fraud, P2P fraud scams, or romance swindling.
The threat is further complicated by the spread of “money mules.” Money mules transfer or move illegally acquired funds to make tracing harder, often using fake identities to avoid detection. In the past, money mules were real people. But now, bad actors in transnational organized crime rings who need to launder millions of dollars create synthetic identities to make money mules they control to facilitate moving illegal money.
The largest enterprises and government agencies stop synthetic identity fraud with Socure’s multi-layered approach which correlates PII, events, and behaviors across businesses and locations using real-time and historical data, velocity intelligence, entity resolution, and link analysis.
Sigma Synthetic Fraud v4 enhancements include:
Innovative email risk enhancements: Email tumbling, or when people create “alias” email addresses by adding punctuation marks like periods between letters, often indicates ill intent. Sigma Synthetic Fraud v4 detects tumbling techniques that are commonly used to commit synthetic fraud, so customers can block the bad actors behind them.
Unparalleled consortium data including feedback: Bringing together a network (Socure Risk Insights Network) of 1,900+ of the world’s largest organizations that span diverse industries and government agencies allows Socure to identify multiple identity elements across the consortium and continually optimize machine learning algorithms to drive the highest accuracy in the market. Bolstered by over 150 million rows of outcomes in the past year, Socure’s database now totals two billion known good and bad identities.
Human-in-the-loop machine learning: Socure fraud investigators provide clean, corrected, and properly classified fraud labels for unlabeled or mislabeled raw data. The labeled data, based on actual synthetic incidents and patterns, becomes training data. Thus, the model is trained to think like a fraudster and applies this intelligence to become smarter at detecting evolving synthetic threats. This unique machine-human intelligence can be used to identify synthetic identities at onboarding, and account changes, and uncover “sleepers” hiding within portfolios.
Real-time fraud attack detection: Socure’s velocity engine tracks how often someone’s personal information is used in applications, as well as how often that information is linked to other data across the Socure Risk Insights Network. Analyzing all of this data on a large scale can help predict fraud attacks before they happen.
Embedded link analysis: Link analysis searches tens of thousands of correlations between an entity’s name, address, email address, phone number, SSN, DOB, IP address, and device intelligence to track fictitious identities across the Socure Risk Insights Network. For example, suppose a bad actor creates accounts using different names or SSNs but uses the same email address, phone number, or physical address. In that case, link analysis will quickly identify these linked fraudulent accounts.