Transferring from guide to automated doc processing has been one of many nice success tales of AI, eradicating numerous hours of repetitive, error-prone guide labor, and releasing folks as much as concentrate on delivering actual worth. Google Cloud’s Doc AI has been a key a part of this success.
However new methods of taking in paperwork can expose organizations to numerous ranges of danger and create new alternatives for fraud to happen, notably in high-risk industries like monetary providers. Know Your Buyer (KYC) and anti-money laundering (AML) laws require these establishments to gather and course of paperwork to know their purchasers higher. Automating these processes with out sufficient controls can as a substitute improve their danger.
The dangers for FinTechs are specifically very excessive, as introducing operational efficiencies like automation into their group is often a key organizational objective — but it surely additionally creates alternatives for varied forms of nefarious actions. A great instance is organized criminals creating faux accounts which might result in fraud and cash laundering actions. For this reason fraud resiliency is among the largest challenges for FinTech Product and Operational groups, notably within the automation of buyer onboarding.
Handbook verification is incapable of figuring out most forgeries in digital paperwork since they depart little-to-no seen traces. After processing tens of hundreds of thousands of paperwork, Resistant AI discovered that as much as 17% of digital financial institution statements used for mortgage functions, account openings, and different functions worldwide have been tampered with, and 15% of firm registration certificates submitted worldwide when opening company financial institution accounts are faux. As well as, serial fraud assaults which depend on easy-to-find doc templates present an affordable technique to scale assaults past human controls and create armies of throwaway accounts that can be utilized as artificial cash mules.
That’s why Resistant AI developed Doc Forensics: an easy-to-deploy API that may automate doc fraud checks to safe any workflow constructed with Google Cloud’s Doc AI.
Automated doc fraud checks
The core objective of Resistant AI’s Doc Forensics is to supply a fast determination to simply accept, deny, or escalate a digital doc for added assessment. The submitted doc is certified, categorised, and verified inside seconds to hurry up the assessment course of, whether or not in PDF or picture format, to make sure a seamless and simple expertise.
-
High quality: Are you able to learn the doc? This step checks the paperwork to make sure they’re of ample high quality to be processed for Optical Character Recognition (OCR), and aren’t hiding fraud behind digicam flashes, low decision, picture blurs or extra. Causes of rejection can be utilized as suggestions for purchasers to submit a brand new model.
-
Classification: Is it the correct of doc? This step ensures you solely absorb paperwork that meet your acceptance standards. You shouldn’t be taking in and processing ID playing cards if all you’re searching for are utility payments.
-
Verification: Do you have to belief it? This step checks the doc over 500 methods for indicators of tampering. After particular person evaluation, it is going to evaluate all submitted paperwork — the whole lot from metadata to the colour temperature and scene composition — to search out patterns of doc reuse, template farms, and mass fraud makes an attempt throughout accounts.
Collectively, these three steps ensure human intervention isn’t wasted on clear fraud circumstances which must be declined, or on paperwork that present no indicators of tampering and which must be robotically processed by Google Cloud’s Doc AI extraction. When a doc is escalated for assessment, it’s displayed in a consumer interface (UI) which makes each single discovering that decided the decision accessible to the reviewer in order that they spend much less time making a call.
