Behind the scenes of Jumio’s industry-leading AI-powered identity verification solutions are a number of cutting-edge technologies. We continually invest in these technologies to keep our solutions at the forefront of fraud detection and UX.
The beating heart of Jumio’s identity verification process is AI and machine learning. We leverage the power and insights from more than 200 million verifications to develop and train robust machine learning models — models that have dramatically improved our identity verification accuracy and verification speeds over the last year. In fact, 90% of Jumio’s identity verification process has been automated through AI and machine learning. We leverage AI in a number of ways, including:
- Automatic ID Recognition: Machine learning speeds up the processing by automatically comparing the submitted ID against known ID templates (3500+ types supported). This helps minimize the response time for a better user experience.
- Data Extraction: Jumio uses deep learning models to extract key data from ID documents.
- Fraud Detection: Deep learning helps us understand the unique fonts, pictures and security features of thousands of government-issued ID types, making it easier to spot characteristics of fraudulent IDs.
- Risk Scoring: Jumio powers machine learning to assign risk levels based on known risk factors. Riskier transactions are subject to augmented intelligence (human-in-the-loop via trained verification experts).
- Selfie-ID Face Matching: AI helps detect the similarity between selfie and photo (government-issued) ID. This translates into fewer false negatives and more good users successfully onboarded.
Jumio uses OCR to extract important data from an ID document, such as a driver’s license or passport. OCR was originally intended for reading black text against a white background often using a flatbed scanner — not for extracting key data fields from ID documents using small fonts and different colored backgrounds that may include holograms, watermarks and printing on glossy surfaces. Powered by AI, Jumio’s OCR engine overcomes the limitations of traditional OCR and is capable of highly accurate data extraction — data that can be used to ping third-party databases or to verify a person’s age.
So when OCR is applied to data extraction from pictures of ID documents it needs a little help from AI to accurately extract key data fields such as name, address, document number and date of birth. When people take pictures of their ID documents using their smartphones or webcams, these images usually need to be de-skewed if the image was not aligned properly and reoriented so that the OCR technology can properly extract the data. Jumio also exploits the power of AI to recognize the patterns that characterize a specific ID type, and this can be challenging given the large number ID subtypes (e.g. some printed in landscape, some in portrait mode).
Jumio’s biometric technology enables us to verify that the ID document truly belongs to the person making the transaction. Biometrics adds a critical layer of protection against stolen IDs and impersonation attacks.
It’s a simple and familiar task for your users — all they have to do is take a selfie. In the process of taking the selfie, Jumio near-instantly captures a short video containing hundreds of still shots. Jumio’s AI automatically selects the best image to perform advanced facial scanning. We compare the user’s facial biometrics to the photo on the ID document, and generate two scores: one for validity and one for facial similarity. The validity score tells whether the selfie is a valid, live selfie — not a prerecorded video, bot or deepfake. The similarity score, which is also powered by AI, indicates the confidence level that the image in the selfie matches the photo in the identity document. If the similarity score is too low, we alert you to the possibility of impersonation fraud.
Thanks to dozens of massive, broad-scale data breaches and the resulting rise of the dark web, legacy verification and authentication methods have been compromised. Deepfakes, bots and advanced spoofing attacks have made certified liveness detection a must-have feature in any biometric-based verification solution. Jumio integrates certified 3D liveness detection from FaceTec to thwart the many well-documented vulnerabilities in 2D liveness detection methods which render them susceptible to spoofing.
Jumio steers clear of gesture-based gimmicks, like asking a user to blink or speak a random passcode, as these techniques add friction to the experience, and are also easily fooled by basic spoofing tactics. In fact, Jumio is the only identity verification solution which leverages certified liveness detection — the first and only biometric technology to achieve Level-1 and Level-2 certification for NIST iBeta Presentation Attack Detection (PAD) testing.
Let a Jumio expert show you how easy it can be to integrate our suite of verification solutions to your website or mobile experience. Start by requesting more information and we’ll be in touch shortly.REQUEST MORE INFORMATION