Sponsored by ?

This article was paid for by a contributing third party.More Information.

Risk Technology Awards 2019: IBM

RRERTA19-IBM

Financial crime product of the year
Innovative vendor of the year

Austin Wells, IBM
Austin Wells, IBM

IBM has taken a cognitive approach to identifying and stopping payments fraud. By combining human expertise with artificial intelligence (AI) and machine learning on an open platform, it is helping financial services institutions adapt faster to changing threats. 

IBM Safer Payments protects all cashless payment types and interaction channels. Its payments fraud-detection engine is proven in production to process thousands of transactions per second with less than five milliseconds latency and extremely low false positives. The cognitive approach and in-memory processing allow institutions to rapidly identify and adapt to changing fraud threats by profiling the behaviour of every customer and their associated entities for all transactions across all relevant channels in real time. 

The system uses AI to act as a ‘virtual analyst’ to assist human experts to find optimal fraud defences. In profiling entity behaviour, the system can make use of channel-specific intelligence, such as cyber fraud management solutions, for a more holistic approach to identifying risky patterns. With a flexible and extensible data model, it can use various data inputs, including financial transactions, non-monetary events, and user and device authentication information from sources such as IBM Trusteer and others. 

Safer Payments provides real-time virtual simulations on current data that allow for fraud analysts to create as many ‘challengers’ to the currently running ‘champion’ profile, model or rule configuration as required without impacting the production environment. This virtual approach is considerably more efficient than taking historical data from production and moving it to a separate analytical test environment. For each virtual simulation of each challenger, any number of attributes representing the likes of regions, transaction types or customer classes can be defined and analysed in parallel, enabling quick and interactive response even on hundreds of millions of payment transaction records. 

Safer Payments is an open data science platform that allows users to export and import models in the PMML open industry model markup language and take in models or feature extractions as Python code. Institutions can use any open-source or vendor machine learning and AI platform or product. Externally trained scoring models can be transferred to Safer Payments and run inside the system, allowing them to be used in any combination with other model components using the platform’s virtual simulation ‘sandboxes’. The system will also provide training and verification data for models developed in other systems. 

IBM has purpose-built Safer Payments for Tier 1 global payments services and infrastructure providers and can integrate with a firm’s payments, security, authentication, case management and other systems via an open data architecture and high-speed processing. The solution can be run on-premise or in a cloud environment. It can be used standalone, or in conjunction with other analytical solutions, such as IBM regulatory and cyber security products. For example, it will integrate with IBM Trusteer security software, as well as the IBM Financial Crimes Insight suite for anti-money laundering, know your customer, employee conduct and financial crime investigation management. The system is certified to the Payment Card Industry Payment Application Data Security Standard and is configured for high reliability.

Safer Payments currently protects more than 600 million customer accounts and nearly 5 million merchants, for a total in excess of 70 billion transactions per year. Users include Stet, the French national payments switch, which processes more than 90% of transactions in France, and Indue, an Australian payments provider that provides payment processing and financial crime intervention on behalf of small and mid-sized institutions in Australia. 

Profiling behaviour across channels and rapidly adapting profile variables, rules and models using a holistic picture of the customer or other entity are critical to an effective and efficient response to financial crime. The IBM Safer Payments dynamic fraud risk engine allows financial institutions to adapt to changing patterns of attack, integrate multiple contextual data sources, iterate new profile variables quickly, apply machine learning and adjust models
on demand.

Austin Wells, Watson financial crimes offering manager at IBM, says: “The growth of real-time, mobile and peer-to-peer payments has led to a corresponding increase in fraud, meaning financial institutions, payment processors, issuers and acquirers need to more accurately control losses without negatively impacting customer adoption or experience. IBM Safer Payments helps organisations make better decisions, faster, by combining transactional, behavioural and session or device data for greater context and using AI, machine learning and advanced analytics for greater performance and scale. In addition, its highly customisable, open data science platform helps data science teams easily create, modify, test and implement custom detection models to derive greater insight out of customer data.”

Judges’ comments

“Impressive commitment to open source and the use of advanced technologies in a targeted way and not just for technology’s sake.”  

“Real-time payment is quickly becoming a reality, and banks are going to need solutions to deal with financial crime that are agile and within their control to implement if they hope to keep up with the threat actors in this space.”

IBM brings open-source and simulation capability to the fraud model risk space that can reduce time to develop and prove new models before implementing. It also creates the transparency necessary to document and defend model parameters to regulatory agencies, which can often be a challenge.”

 

Read more articles from the 2019 Risk Technology Awards

  • LinkedIn  
  • Save this article
  • Print this page  

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: