In light of the COVID-19 pandemic, those predictions remain on course – and in fact have become even more relevant.
The reason is that with the volume of immediate payments – forecast in our “Prime Time for Real-Time” report – exploding due to pandemic-driven social restrictions, so too have the volume and variety of possible attack vectors. Criminals, as ever, have been quick to take advantage, and reports have emerged of increased targeting of both FIs and their customers.
The immediacy of real-time transactions significantly reduces the time available to detect fraud and the recoverability of funds when fraud does slip through the net. As such, there’s a greater need than initially predicted for machine learning and intelligence sharing to help FIs keep one step ahead of fraudsters.
Strength in numbers
Fighting fraud alone rarely pays off. That is why, despite their fiercely competitive nature, there’s a long-standing tradition of co-operation between FIs.
Through improved collaboration, FIs can create a community where real-time information on emerging risks is freely shared between members, including central infrastructure (CI) owners. The shared intelligence capabilities that are part of ACI’s model generator feature are a great example of this – fraud threats can be quickly identified and the community alerted to emerging fraud characteristics, meaning they can be incorporated into their machine learning models more quickly and cost-effectively than ever before.
An added benefit of the approach is that CIs can move beyond a mere supervisory role and instead become actively involved in coordinating or facilitating the community’s fraud defenses. Without interfering in the decisions that members make regarding their own individual fraud detection strategies, a community facilitates end-to-end communication between FIs and CIs to increase transparency, without compromising data-sharing compliance.
Overall, this collective front is better able to respond to new and emerging risks and prevent them from becoming endemic threats.
Compliant, community information sharing
Compliant information sharing is achieved by enabling the community to share, in real time and in meta-data format, their fraud models in their composite features, along with key performance data supporting their efficacy. Automatically stripping the metadata of any identifiable information resolves the burden and regulatory risks around attempting to extrapolate and submit data externally. Thus, the community can share more data at lower risk.
While the community’s central body, perhaps a CI or governing body, controls quality and oversight by pre-aggregating the data to form consistency in data and flow across members, it does not decide which models the group adopts – the community does. Furthermore, members can adopt, adapt or combine features with their own models however they see fit and with no limit on the number of models that can run concurrently. In this way, participants can integrate proven model features into their own customized adaptive machine learning strategies. This drives unparalleled access to all the information needed to assess transaction risk levels.
Agility and adaptability are crucial in a real-time world
The need to build out, deploy and constantly adapt advanced, predictive machine learning models is more in demand now than ever before.
Financial institutions need solutions that reduce their reliance on specialized resources and allow them to adopt a business-led machine learning strategy to match today’s fast-paced, 24/7 fight against fraud. That means implementing solutions that support the complete model development process, allowing for easy access to examine and analyze data, calculate scenarios and document key modeling steps.
To learn more about the shared intelligence capabilities within ACI’s fraud management solution, download our eBook “Expanding horizons of fraud detection.”
ACI Fraud Management for Banking 
