Dispersed Financial Fraud Enhanced Detection

Protect clients from fraud with the help of their mobile phones

IBM DEFFEND takes financial fraud detection to the next level, offering banks and credit card companies added protection that is built into their customers' personal smartphone devices.

Most of us keep our mobile device within reach, either on us or very close by. Why not use this fact to help verify financial transactions? If you like to travel and shop in new places, classic fraud detection solutions won’t raise an alert if you buy a new coat in Mexico. But if a client’s phone shows that she’s still in New York, DEFFEND allows you to notify her and ask her to verify the transaction.

We examine calls, location, browsing history, installed apps and logs, and more—to understand what is ‘normal’ and what might be suspicious. Your bank doesn’t know where you are every minute of the day, but your phone does!

DEFFEND combines existing, centralized solutions with smartphone-based analytics so that private information is never shared, copied, moved, or exposed.

DEFFEND is different from classic centralized fraud detection systems that verify transactions only at the bank or parent organization level. We complement centralized information and analysis with data on the user’s device to protect privacy.

Think of DEFFEND as a booster shot for your fraud detection system, adding another layer of protection.

How it works

Our system uses sophisticated machine learning algorithms to analyze what represents “normal” behavior and what does not. By learning about the user’s regular behavior and being aware of the current context, we can identify certain seemingly innocent activities as suspicious ones.

Our system uses behavior on the mobile phone to do the following:

  • Create a user profile to understand what is usual and what is not.
  • Generate user behavior analytics (UBA) by learning what key-in method, speed, and habits are used on the phone.
  • Understand the context of the user, whether by location, activity, or movement.

All of the above - learning user behavior, UBA, and context awareness -- are performed exclusively on the user’s device. There’s no need to send the data to the transaction center or share it with any third party. Once the above steps are carried out, machine learning models use the context features alongside other transaction characteristics to produce state-of-the-art prediction capabilities.

Adding muscle to standard fraud detection

  • Integrates into existing systems – no switching
  • Conducts verification without speaking to the user

What about privacy?

With DEFFEND in place, users’ personal data is used only for lawful fraud detection, creating a win-win situation. User information never leaves the smartphone when verifying data.

Legislation and privacy regulations are becoming stricter when it comes to collecting and processing personal data. Most now require the informed consent of individuals to process their data for specific purposes. But IT solutions that can address these privacy issues are still lacking.

DEFFEND automatically deploys the enforcement and audit of privacy policies at multiple levels, while collecting and analyzing data for the purpose of fraud detection.

Users who want to benefit from DEFFEND can opt in, and specify how they permit their data to be used—only if there is advanced suspicion of fraud, only for a limited amount of time, and so forth.


Contact Us

Contact us if you’d like to be part of a pilot or hear more about this solution