The Fraud Lab
Identity fraud is big business. It’s created an entire industry complete with accessible tooling, training, and powerful techniques for professional fraudsters. To keep our customers one step ahead, we've created our in-house Fraud Lab to train and test our solutions using those same professional attack patterns.
Join them to
Fraud Lab allows us to continually test our solution with data that mirrors real-world conditions.
Emerging attack patterns
We can replicate emerging attack patterns as soon as we encounter something new to close vulnerabilities fast.
Imagining the future
Our experts think like fraudsters, imagining what patterns and tools we’re likely to encounter next — keeping our products ahead of new technologies, like deep fakes.
Identity Fraud Report 2024
Get our analysis of the latest identity fraud trends, attack patterns, and prevention tips, to help you build a data-driven defense. We dive into deep fakes, synthetic identities, and what industries and regions are at the highest risk.
How does the Fraud Lab keep us ahead?
Analysis and proactive benchmarking
Our experts analyze every fraud case to close the gaps in our solutions, and more effectively improve our performance. The Fraud Lab helps us set our roadmap and stay one step ahead of the latest fraud.
You can’t train AI without data — and in the world of fraud, data can be hard to come by. Our proprietary tools are able to create the thousands of fraud samples needed to replicate attacks, and create effective datasets to train our Atlas AI.
Solutions powered by Fraud Lab
Our award-winning document verification supports over 2500 documents worldwide. Every ID, from every country, in it’s many multiple iterations has unique vulnerabilities. The Fraud Lab enables us to test individual document types, and reverse engineer new attack vectors we’re seeing in the real world, so we can react quickly and retrain Atlas’ 10,000+ micro-models to catch emerging techniques.
Our AI-powered biometric analysis contends with 2D and 3D masks, photos of screens, printouts, and most recently, deep fakes. Fraud Lab works hand in hand with our product and research teams to train our AI models and inform our roadmap.
That’s how we created Onfido Motion — our latest and most secure biometric verification. It’s 100% automated, iBeta PAD Level 2 compliant, and returns 95% of verifications in seconds — and it was designed with data both collected in the field and generated in the Fraud Lab.
Fraud is becoming increasingly sophisticated and organized. The Fraud Lab tracks the activity of fraud rings to understand how they operate. We frequently see these rings attack at scale with automated tooling, using hundreds of synthetic identities — often with minor tweaks to document details or biometrics.
We created Known Faces and Repeat Attempts in response. They compare identity documents and biometrics against previously seen examples of fraud, and flag if they’ve been seen before. They’re both fully automated and allow us to identify organized fraud at zero additional friction to good users.
Onfido Device Intelligence detects sophisticated fraud by analyzing information about the state of a user’s device and network. It considers data points such as IP address, geolocation, and device metadata to find inconsistencies that point to detect repeat fraudsters and fraud rings — all at zero additional friction to users.