How AI Is Influencing Ethiopian Banks

Ethiopian banks lost 1.8 billion birr to fraud over a five-year period, before digital payment adoption accelerated. That figure matters because the fraud that drove it was largely the kind banks have always known how to look for: internal. Privileged users abusing system access, employees exploit institutional knowledge to target dormant accounts and back-office transactions that attract little routine scrutiny. Detectable, if you know where to look.

What has changed is everything around it. Account takeover through SIM swap gave fraudsters easy access to mobile banking accounts until banks and telcos began closing that loophole — at which point fraudsters simply pivoted to calling customers directly and manipulating them into authorizing transfers themselves. Cloned websites, synthetic identity, automated scams running at a scale no human fraud ring could sustain. The external threat has become organized, fast, and increasingly AI-assisted.

At the same time, leading Ethiopian banks are responding to sharper regulatory expectations: AML obligations, suspicious transaction reporting within 24 hours, risk-based supervision, and the Fayda biometric eKYC rollout mandated nationwide by December 2026. The compliance pressure and the fraud pressure are arriving simultaneously, and they require the same thing: a compliant, real-time fraud detection and AML monitoring system that catches financial crime before the money moves.

What Changes When Detection Is Built On AI

Rule-based systems flag what they already know to look for. The problem is that the most dangerous fraud does not match a known pattern. It looks almost normal, deviating in one or two signals that a rule would never catch. A first-time transfer to an unfamiliar beneficiary, a login from an unusual device at an unusual hour, or a payment that matches the customer’s history in every way except one.

AI-powered behavioural analytics learns what normal looks like for each customer, each account, each payment type — and surfaces those deviations in real time. That is the shift from catching fraud after it has happened to stop it before the money moves. It is also what makes continuous identity verification possible: not a static check at onboarding, but an ongoing understanding of how a customer behaves, so that when a SIM swap hands a fraudster someone’s credentials, the behavior that follows flags it immediately.

And when an alert does fire, investigation cannot be an afterthought. A fraud team needs to know where the payment originated, which channels it moved through, and whether similar patterns are active elsewhere — before they open the case. That requires tooling that consolidates alerts, tracks investigator actions, and preserves the evidential links that regulators will ask for. Detection and investigation have to be built together.

From NetGuardians To Vyntra

Many of you know us as NetGuardians, where we spent nearly two decades helping banks detect fraud through behavioral analytics — deployed across more than 130 financial institutions in over 60 countries, including three banks in Ethiopia and leading banks across East Africa. That experience is the foundation of everything we do.

In 2025, we merged with Intix to create Vyntra, bringing fraud detection and AML monitoring together with end-to-end payment visibility in a single platform. The merger was a direct response to fraud that moves across systems while institutions aren’t equipped to see the full transaction journey.

In 2027, Chartis ranked Vyntra in the RiskTech100^® and recognised us with category awards for AML and anti-fraud solutions.

Come find us at CBE’s 3rd Cybersecurity Awareness Month Conference.

Vyntra will be in Addis Ababa on 29-30 October 2026 to showcase our FinCrime capabilities. On Day 2, I will be on the panel Who Is Actually Opening This Account?, exploring how AI-powered behavioral analytics is reshaping how Ethiopian banks approach account security.

This article was written by Jonathan Somers, Regional Sales Director Middle East & Africa, Vyntra