AI fraud detection in Indonesian financial services: what our survey found
Article
Where is AI already at work in Indonesian banks and lenders? Fraud detection is near the top of the list. 42% of the people we surveyed inside them use or test AI for fraud detection, level with reporting and analytics and behind only process automation (58%). This article covers what our survey found, the scale of the scam problem fraud teams are dealing with, what a fraud model needs from your core and how Oradian connects the two.
Only got five minutes? Here are the key takeaways of this article:
42% of Indonesian respondents use or test AI for fraud detection. Risk and decision-making AI, which covers fraud detection, credit decisions, compliance and AML, is used or tested by 67%.
Scam losses reported to Indonesia’s Anti-Scam Centre reached IDR 9.1 trillion between November 2024 and mid-January 2026.
A fraud model needs each transaction at the moment it happens, plus monitoring and explanations that match how much its decisions affect customers.
Real-time events, governed data access, your own rules and a full audit trail are what fraud teams need from the core.
Get the full findings in From AI adoption to AI at scale. You can also book a call with vanda.jirasek@oradian.com to talk it through with our Jakarta team.
Want to find out more first? Read the rest of this.
What our survey found
We asked where respondents’ institutions use or test AI today. Process automation came first (58%), with fraud detection and reporting and analytics close behind (42% each). Customer service and credit decisions followed (36% each), then personalised offers (30%), compliance and AML (30%) and collections (24%). On average, respondents named three areas.
Grouped together, risk and decision-making AI (fraud detection, credit decisions, compliance and AML) is used or tested by 67% of respondents. Respondents further along use it more often for compliance and AML (40% against 15% at an early stage) and for credit decisions (45% against 23%).
The scale of the problem
Scam losses reported to Indonesia’s Anti-Scam Centre reached IDR 9.1 trillion between November 2024 and mid-January 2026, across 432,637 reports. By June 2026 the centre had handled more than 608,000 reports and more than 557,000 accounts had been blocked.
OJK’s Regulation 12 of 2024 sets out anti-fraud strategy requirements for financial services institutions. When OJK launched its AI governance guidance for Indonesian banking, its launch announcement listed fraud prevention among the areas where AI use would keep growing.
What a fraud model needs from your core
Current data: each transaction and account change at the moment it happens, so suspicious activity can be assessed in time to act.
A way to act: the fraud engine’s decision has to reach the systems that can stop or flag a payment.
Data to investigate: analysts need complete history to find patterns without slowing the live system.
Controls you can explain: monitoring and explanations that match how much the model affects customers and the institution.
How Oradian connects fraud tools to the core
Real-time events: Kafka and HTTP webhooks publish each transaction and account change as it happens, so your fraud engine sees it in time to act.
Stable APIs: 750+ native, versioned API endpoints let the fraud engine’s decision flow back into the workflow.
Database Access: a governed, read-only copy of your production database gives fraud analysts complete, current data to investigate patterns without touching the live system.
Custom Code: your own fraud and credit rules run inside the core, versioned and auditable.
Audit trail: every product change, partner connection and automated step is logged and ready for regulators.
Locality: Oradian’s own team in Jakarta implements the platform and works directly with yours.
Give your fraud team current data
Fraud detection is one of the most common uses of AI among our Indonesian respondents. A fraud model can only act on what reaches it in time, so the platform it reads from matters. Read the full findings in From AI adoption to AI at scale and book a call by emailing vanda.jirasek@oradian.com to see how Oradian connects your fraud tools to the core before the end of the year.