The ai Corporation (ai) has developed a range of self-service machine-learning products designed to help banks fight Automated Clearing House (ACH) fraud. ai’s SmartSuite enables clients to put their own rule sets into any fraud platform, including ai’s rules engine RiskNet.

Machine learning gives banks the opportunity to automate their fraud prevention, including ACH fraud regardless of the penetration method, like ‘man in the middle’ fraud and/or social engineering. The ACH network processes payments across individuals and businesses among banks, so only the balances are actually paid in cash.

ai’s SmartSuite has a total of six products, one of them being SmartScore. SmartScore can recognise patterns and trends in fraud by creating neural models specific to a particular fraud type, using ai and machine learning.t provides transaction risk scores, which can be coupled with user-defined rules and parameter-mapping to decide the course of action, making sure that RiskNet® or any third-party fraud platform users are not reviewing unnecessary alerts.

It then provides transaction risk scores, which can be coupled with user-defined rules and parameter-mapping to decide the course of action, making sure that RiskNet or any third-party fraud platform users are not reviewing unnecessary alerts.

ai also offers data visualisation tools that allow for a granular level of interrogation, highlighting potential risks hidden in large amounts of data. The company aims to offer self-service, automated, and highly scalable products in a SaaS environment.

Academic involvement

ai partnered with Southampton University and Leuven University in Belgium to form a Research Council to investigate the benefits of using true machine learning in fraud prevention.

“One of the team’s priorities will be the adoption of true machine-deep learning. This is the movement away from supervised to an unsupervised self-learning system, with the purpose of fully automating fraud prevention activities for all payment types,” said Tom Myles, Chair of the Research Council and CTO at ai.

“Alongside our partners at Southampton and Leuven Universities, our focus will be to build on our view that fraud prevention platforms are capable of more than detecting fraud. We want to look at ways we can leverage ai’s technology to create solutions that can automate all manner of processes. Not just fraud prevention, but other areas of the business such as credit scoring, credit monitoring, gateway switching and interchange optimisation. Ultimately, ai’s platform is a reusable technology asset that has a wider business application as an end-to- end decision engine.”

ACH payments attracting fraud

By September 2016, credit-based ACH payments are now being settled within the same day, including direct deposit, payroll, person-to-person and vendor payments. Debit-based payments will follow this year.

Consequently, the fraud exposure has increased due to the reduced time to analyse and process suspicious activity. ai also reports an increase in fraud attempts as a result.

On top of that, scammers just need a checking account number and a bank routing number to perform an ACH fraud, which they usually obtain via phishing emails that include malware.

Some types of ACH fraud include payroll fraud and ‘kiting’ – where a criminal takes advantage of the time lag in transactions by juggling funds back and forth between bank accounts at separate banks.

Learning machine learning

Machine learning is one of the talks of the town at the moment, with many institutions developing or implementing these ai capabilities in some way or another.

Last year, Atom deployed the WDS Virtual Agent into its mobile app, a move designed to analyse customer behaviour, problems and solution success rates.

InsurTech platform revenues will reach almost $235 billion globally by 2021. According to Juniper Research, this will be largely driven by machine learning investments enabling insurance providers to personalise products.

Just in the beginning of the year, the Royal Bank of Canada saw an AI expert joining their ranks to boost machine learning processes.

CollectAI is another company that IBS had the opportunity to talk to during MoneyConf 2017 that uses Ai and machine learning in its products. By gathering and reading the data about the way customers respond to debt collections, their software tailors the tone and content of the message as well as the channel utilised to reach out, in order to achieve the most effective and customer-friendly approach to the customer.

What it’s clear is that machine learning is here to stay, and in the next year we will see more and more applications of AI that will change many of the established ways we deal with customers or transactions.

by Henry Vilar
Henry is Junior Reporter at IBS Intelligence, follow him on Twitter or contact him at: