Using data to determine right from wrong… better
What’s black, white, and read all over? If you’ve heard this joke before, you might say a “newspaper.” But the digital world has all but eliminated them. The real answer? Data. When you read it, data is black and white – it’s binary, running the computers, internet, and all digital infrastructure people rely on today. No matter how you format your excel sheet or pump up your numbers, the data won’t lie: the answer to 2 + 2 will always equal 4, not 5.
Let’s pose another question: can data be used to make human’s lives more black and white? We all know that people live in measures of grey – there’s no truly right or wrong answer to moral dilemmas. And in modern times, it’s become increasingly difficult to inform business decisions without using data, analytics, and metrics to measure risks or potential success.
At Mitek Systems -an enterprise company that creates digital tools to help businesses authenticate ID documents- they ask similar questions about data, but in a different way than finding out which companies are most likely to buy their products. Instead, they’re figuring out how data can fight fakes and catch fraudsters better than before.
The fraud landscape is shifting
In 2020, sophisticated fraudsters now have advanced tools to circumvent current biometric fraud detection methods like spoofing device and geo-location data (read Advantages and disadvantages of biometrics blog entry to learn more). They’ve invested a considerable amount of time and effort shoring up valuable fraud profiles called “synthetic identities” that masquerade as legitimate users and beat the best models. The fraudsters that get through to targets bide their time till the moment of maximum advantage before suddenly busting out and taking as much as possible.
In Mitek’s latest featured white paper “Fraud trends and tectonics,” VP of Product Sanjay Gupta cites that online payment fraud for businesses in eCommerce, banking services, money transfer and airline ticketing could reach $200 billion in cumulative losses by 2024.
Mitek Fraud series: Biometrics and Fraud in the COVID era. What's changing? | Identity Verification through the customer journey | The Pandemic blindsided us; will surging digital fraud do the same? | How to fight fraud with data | What is a Deepfake and how does it impact fraud? | Financial services and online marketplaces face shifting fraud landscapes | What is synthetic identity fraud? | What is Account Takeover Fraud
Data now plays an active role fighting fraud
With online fraud quickly becoming a top of mind issue, customers will need to rely on innovative and state of the art solutions. That’s where data comes in.
“From a tech company point of view, people often think about data for use in product reports, informing business insights, and analysis of our tools’ effectiveness,” said Mitek representatives. “At Mitek, we instead ask ourselves how we can use anonymized data in a way that can catch fraudsters that’s compliant with regulations and doesn’t expose a person’s personal identifying information.” Though it seems like a hard practice to balance, Mitek has effectively analyzed hundreds of thousands of anonymized data points captured using Mitek’s Mobile Verify software.
After comparing data from legitimate ID documents to the best fakes Mitek had caught from fraudsters, the company’s identity verification product, Mobile Verify, includes a detection method leveraging anonymous data that effectively helps businesses catch fraudsters. Most impressive, they didn’t have any help or insights from the governments that issue IDs – it was all gleaned from the company’s own due-diligence and analysis.
Data is black and white. And incorrect data can become a signature for both legitimate and fake documents. Mitek uses unique data analysis as a key part of its advanced arsenal to catch both novice and sophisticated fraudsters in thousands of documents they’re tasked with analyzing.
“A novice may not look at data embedded in a document, and, we’re able to tell right away with Mobile Verify,” say Mitek reps. “A more sophisticated fraudster may try to incorporate data into their forgery. But, using our years of experience in the business, highly trained personnel, and state-of-the-art technology, we’re able to stay ahead of fraudster’s own technological improvements and increasing prowess to determine whether or not a document is a fake.”
What should a company do when they’re trying to beat fraud?
Mitek’s recommendation: go active when you must - see the "Synthetic Identity infographic" on why this is important. In very risky situations, request a snapshot of a government-issued hardcopy ID, and take a few seconds to verify the authenticity of the document by comparing the photo on the ID against the user’s selfie. Data analysis can also be run to mitigate fraud with increased certainty.
In the identity verification space, competing tech companies have patented hundreds of ways artificial intelligence, machine learning, and computer vision can pick up document forgeries – flagging the odd placement of text, photoshop markings on a picture, or a missing UV stamp. But a part of the issue with these tools is they won’t dictate, with certainty, whether or not a document is a forgery.
“Right now, artificial intelligence, machine learning, or computer vision will only tell you on a scale if the document is a real or fake. It’s not going to give you that yes or no answer” says Mitek. “If you have a real ID that’s got a scratch on it, the computer vision will give you a response saying ‘this falls in the range of being a fake’ even though it’s legitimate.”
The positive about Mitek’s Mobile Verify is it analyzes data-derived heuristics in combo with the computer vision and AI/ML most businesses now expect from ID document solutions and providers. How is this better? It delivers a stronger answer when providing an authenticity judgement. “Even if a fraudster puts some real effort into creating a forgery, more likely than not they don’t understand how the data in an ID document needs to be constructed and formatted. The difference? We do,” says Mitek.
Mitek continues to evolve its authentication techniques beyond machine learning. Today, ‘big data analytics’ isn’t just a catchy phrase for the San Diego company. It’s a real component that’s powering intelligence in verification software, catching fraudsters, and creating certainty in an uncertain world.