Anatomy of a mule network: patterns across the account lifecycle
A typology of how mule accounts are recruited, activated and exhausted, and the signals at each stage.
ForthcomingTypology
Independent research institute · Hyderabad
Research for a safer digital economy.
The Centre for Financial Crime & AI studies how fraud, financial crime and artificial intelligence are reshaping trust in digital finance, and turns that evidence into practical guidance for the institutions that protect it.
InstitutionsAccountsAnomalous cluster
i.
Money moves in seconds. So does fraud.
Real-time payments, instant credit and open banking have expanded access at extraordinary speed. The controls around them have not always kept pace.
ii.
AI now sits on both sides.
The same models that detect fraud also write the scam message, clone the voice and fake the document. Defenders need evidence, not hype.
iii.
Trust is infrastructure.
When people stop trusting digital payments, inclusion stalls. Protecting that trust deserves the same rigour as building the rails themselves.
What we study
Three pillars, one question: how do we keep digital finance trustworthy?
Our work sits at the intersection of fraud operations, data science and public policy, so that the evidence travels from the research team to the people who make decisions.
01
Financial Crime & Fraud
How scams, mule networks and payment fraud actually work, and which interventions measurably reduce harm.
- Scam typologies
- Mule networks
- Payment fraud
- AML
Responsible AI
Making AI in financial decisions explainable, fair and well-governed, and preparing for AI-enabled attacks.
- Model risk
- Explainability
- Fairness
- Deepfakes
Digital Trust
Identity, consumer protection and the conditions under which people, especially new users, trust digital financial services.
- Digital identity
- Consumer protection
- Inclusion
- Recovery
How we work
Built for credibility, not for a sales pitch.
CFCAI is vendor-neutral. We don't sell products, and our findings are not for sale. That's what makes them useful to banks, regulators and researchers alike.
01Independent
Research questions and conclusions are set by CFCAI. Funders and partners never get editorial control.
02Evidence-led
Claims are grounded in data, case analysis and transparent methods that others can examine and challenge.
03Practitioner-grounded
Our team has run fraud and risk functions inside large institutions. We write for people who have to act on it.
04Open by default
We publish what we learn wherever confidentiality allows, because collective defence beats isolated defence.
Research agenda
Questions we're working on.
A sample of the open problems shaping our current programme. If your organisation is wrestling with one of them, we'd like to hear from you.
- Which friction at the moment of payment actually stops an authorised-push-payment scam, and which only annoys the customer?Financial crime
- How can institutions detect money-mule networks earlier without sharing raw customer data?Financial crime
- What does a meaningful explanation of an automated fraud or credit decision look like to the person affected?Responsible AI
- How should verification processes adapt when voice, video and documents can be convincingly synthesised?Responsible AI
- What helps first-time digital users recover, financially and in confidence, after being defrauded?Digital trust
Insights
From the research desk.
Working papers, practitioner briefs and policy notes. Our first publications are in preparation.
Explainability for fraud models: what reviewers, customers and regulators each need
Three audiences, three different definitions of "explained", and a practical template for each.
ForthcomingBrief
After the scam: designing recovery journeys that restore trust
What victims experience after reporting fraud, and where institutions can reduce secondary harm.
ForthcomingPolicy
Have a problem worth studying?
Bring us the question your organisation can't answer alone. We'll tell you honestly whether it's a fit for independent research.