Share fraud signals, never the data.
Fraud crosses sectors; the evidence doesn't. The bank, the webshop and the police each see a piece of the pattern, and none of them may share it. Dutch government agencies are piloting mintBlue to match those pieces without exposing the data behind them, producing legally binding evidence under eIDAS.
Piloted with Dutch government agencies for fraud detection and identity verification
mintBlue turned our nightmare of invoice exchanges into a dream of automation. Now we strive to make taxation less of a headache for everyone involved.
Claire Arens
Innovation & Strategy, Netherlands Tax Administration
6 million invoices annually validated and processed automatically. No manual reconciliation. No disputes over what was agreed.
Sebastian Toet
Solutions Architect, VISMA | Yuki
Real-time, verifiable carbon tracking across our entire supply chain without exposing sensitive supplier data.
Pauline Van Ostaeyen
Cofounder, Dockflow
Three verification problems, one infrastructure
Identity fraud, cross-authority intelligence, and document integrity all require the same foundation: privacy-enhancing technologies that prove what happened, when, and by whom, without exposing the data behind it.
HOW IT WORKS
From raw data to verified intelligence in four steps
Data stays where it is. Verification happens across organisations without centralisation.
- 01
Data stays where it is
Each participating organisation keeps full control of its own data, with no central database or data lake. Records remain in the systems they already use.
- 02
Records are hashed and sealed
Relevant data points are converted to cryptographic hashes at each organisation. The original data never leaves. Only the hashes, which cannot be reversed into the original information, are used for matching.
- 03
Hashes are matched across authorities
Hashes from different organisations are compared automatically. When matches exceed a predefined threshold set by the participating authorities, the system flags a potential pattern. No human sees raw data from another organisation during this step.
- 04
Targeted data is shared with proof
Only when the threshold is met, specific data is decrypted and shared with the relevant parties. Every disclosure is logged with a cryptographic audit trail: who accessed what, when, and under which legal basis. The entire chain is legally binding evidence.
Why governments are moving away from centralised platforms and manual processes
Manual processes offer strong privacy by default but zero cross-organisational intelligence. Centralised surveillance platforms offer full visibility but require every participant to hand over data custody to a single vendor. Multi-party computation and clean rooms answer a bounded question on a batch of data: they compute on data nobody can see. mintBlue works in the stream and proves what happened to data through verifiable credentials: who did what, when, and on what basis. Matching runs on signed markers over a shared record, with no private compute in the middle.
Distributed verification removes that trade-off.
Manual Processes & Centralised Platforms
- Fraud pattern detection limited to what one authority can see in its own data
- Privacy protection either too high (preventing legitimate sharing) or too low (vendor sees everything)
- Legal admissibility disputed. Evidence trails are incomplete and often contested.
- Does not scale. More authorities means more bilateral agreements and more overhead.
mintBlue
- Cross-authority matching via blind hashing. Patterns detected without exposing raw data.
- Privacy built in. Data is never exposed until a verified threshold is exceeded. GDPR-compliant.
- Legally binding across the EU under eIDAS. Cryptographic seals on every record.
- Scales with participating organisations and data volume. No per-seat licensing.
Share this with your board
Fraud detection that respects the boundaries it crosses
The full story on one page your board can read in a minute.
Two champion one-pagers, fraud signals and photo provenance, in one email. · Schedule briefing
Share this with your board
Fraud detection that respects the boundaries it crosses
The full story on one page your board can read in a minute.
Two champion one-pagers, fraud signals and photo provenance, in one email. · Schedule briefing