How AI is Being Used to Predict Fraud

Financial institutions are always looking for new ways to protect customers. Artificial Intelligence is one technology that could help stop fraud.

How AI is Being Used to Predict Fraud

Financial institutions are always looking for new ways to protect customers. Artificial Intelligence is one technology that could help stop fraud.

Written by:

Ivan Pointer

Read time:

3 mins

Category:

Artificial Intelligence

In theory, financial fraud should be easy to spot. Most fraud attempts involve transactions that are out-of-the-ordinary when compared to an account’s usual activity.

This might mean money spent in a new location, a high amount of money sent to a new account, or purchases on unusual items.

However, the reality is that financial institutions sometimes fail to spot fraud. This is unsurprising considering the number of wire transfers and credit card transactions that are made daily.

Fraud can have disastrous consequences for the people targeted. Victims find themselves swindled out of a large amount of money that can be difficult to get back.

Because of this, financial institutions are always looking for new ways to protect customers. One such technology that promises to make fraud-detection easier is Artificial Intelligence (AI).

Why AI is Better at Fraud Detection Than Existing Solutions

Banks have had methods in place to stop fraud for a long time. Non-AI based solutions generally use a set of predefined rules to spot fraud. These cover factors such as transaction amounts and what money is spent on.

The problem is they aren’t always 100 percent effective. Not only do they not always spot fraud, but they can also result in a high number of false positives — transactions that get flagged as fraud but aren’t.  

Systems that use AI could increase the effectiveness of fraud detection. They can analyze a high-number of transactions and data points in real-time.  The software can then build up a more accurate picture of what a fraudulent transaction looks like in comparison to a customer’s normal behavior.

This could result in both an increase in the number of fraudulent transactions that get caught and a decrease in false positives.

Here is a look at some organizations already using AI to detect fraud.

Mastercard: Stopping Credit Card Fraud

Back in 2016, Mastercard introduced Decision Intelligence, a solution that uses AI to spot when a payment is fraudulent. At the time, Mastercard said it hoped the system would reduce the number of false declines experienced by its customers.

Mastercard says the AI-based system is able to learn from each transaction that a customer makes. It then uses this data to analyze whether follow up purchases are fraudulent.

Over time, the system is able to detect unusual shopping habits. According to Mastercard, the data used includes customer value segmentation, risk profiling, location, merchant, device data, time of day, and the type of purchase made.

Monzo: Detecting Fraudulent Transfers

Monzo is an online bank that uses AI and machine learning to detect when a deposit made into one of its accounts is unusual.

The company says that this is especially useful to help discover accounts utilized by money mules. These are accounts fraudsters use to move around stolen funds. If a criminal can do this successfully, it can make it difficult for victims to recover their losses.

When Monzo flags a transfer as suspicious, the company blocks the money so that it can’t be accessed until it confirms the transfer was legitimate. This stops criminals from moving money gained through illegal means.

Financial Regulators: Making Money Laundering Less Appealing

Rob Gruppetta, head of the financial crime department of the UK’s Financial Conduct Authority, recently spoke about how AI could be used to discover fraudulent transactions. This could help the organization see when criminals are laundering money.

Mr. Gruppetta said that the organization is moving away from a rules-based approach, towards a data-driven predictive approach.

If it becomes harder for criminals to move stolen money, it could be easier for victims to get their money back. It could also make committing a crime less appealing in the first place.

Darktrace: Early Protection from Fraud Attempts

The above examples focus on financial institutions stopping fraudulent transactions. But, it is also important to stop fraudsters accessing information that could help them deploy a scam.

One way to do this is by protecting your company’s computer network. That is what security solution Darktrace uses AI to do.

According to the company’s chief executive for the EMEA region, Darktrace uses AI to see what a system’s normal activity looks like. Much like Mastercard detecting unusual payments, the software is able to recognize network activity that doesn’t fit the usual pattern.

Hacks found by Darktrace include one that took advantage of a fish tank. The tank had network access to notify someone if the temperature dropped below a certain point. However, an attacker exploited a weakness in the software to send data to Finland.

Darktrace noticed the amount of data being sent by the fish tank was higher than normal, allowing the vulnerability to be fixed.

AI Could Make Fraud Detection More Effective

As AI fraud detection systems improve, it could become harder and harder for fraudsters to commit scams. Still, recent cases show that, despite this potential, we can’t rely on existing AI solutions to protect customers in the real estate industry.

Instead, businesses need to have their own fraud protection systems in place to shield themselves from scams like business email compromise.  Title agencies can take steps to secure all accounts using strong passwords and two-factor authentication.

Additionally, everyone involved in the sales process should be kept up to date about the signs of fraud. If customers know what to look out for, they will be well prepared to spot fraud attempts if they are targeted.

Finally, when it comes to making wire transfers, it is important to have an effective way to authenticate everyone involved in the process. This will ensure that any money sent will go to the correct person.

We believe that CertifID is a great way to authenticate real estate wire transfers. To get a free ten-day trial to our service, sign up here.

Ivan Pointer

Engineering Lead

Ivan is passionate about building great teams, great culture, and great software. Before CertifID, Ivan worked with companies like EDS, HP, NIC, and SailPoint, gaining experience in government, enterprise architecture, and security.