Cross-bank fraud intelligence: industry consortium pilot

Context

4 commercial banks UZ. Each has fraud detection capability, but cross-bank fraud (attacker moving between banks) is invisible. Known patterns: SIM swap → account takeover → cash-out through secondary bank.

Bank CROs joint discussion — shared intelligence framework via industry consortium.

Diagnostic (4 weeks)

Sample analysis:

• ~30% of suspected fraud cases involve cross-bank flows.
• 12% of recovered fraud could have been prevented with early cross-bank signals.
• Regulatory framework for shared fraud intelligence — emerging, not fully formed.

Approach (9 months)

Months 1-3. Clean room architecture + legal framework. Privacy-preserving signal sharing — hashed identifiers, no raw PII.

Months 3-5. Regulator engagement. cbu.uz dialogue for legal framework approval.

Months 5-7. Platform pilot. Each bank publishes fraud signals (hashed customer ID + risk score), consumes others’. Real-time matching.

Months 7-9. Operating routine. Daily signal exchange, monthly cross-bank fraud committee, quarterly metrics review.

Results

After 9 months:

• Cross-bank fraud detection rate +60%.
• Estimated fraud prevention $2.4M annualised across 4 banks.
• False positive rate maintained — strict matching protocols.
• Regulator approval for framework — positive precedent.
• Industry consortium expanded to 7 banks in follow-up phase.

What is critical

CRO leadership across banks — coalition essential.

Privacy-preserving architecture — without compromise on customer data.

Regulator engagement — key for legal framework.

Clear value sharing — all participants benefit equally.

What would not have worked

Bilateral data sharing — too complex, no scale.

Raw data sharing — regulator violation.

Industry-wide without regulator buy-in — legal risk.

Related

• /en/architecture/banking-data-clean-room/ — clean room
• /en/solutions/banking-fraud-platform/ — fraud platform
• /en/insights/banking-fraud-trends-uz/ — fraud trends
• /en/expertise/banking-fraud-discipline/ — fraud expertise