Detect duplicate invoice financing across banks
Sector: Tertiary sector · Industry: Banking and financial services · Organisation: e& enterprise (vormals Etisalat Digital) · Maturity level: Production
Telecommunications group e& enterprise, formerly Etisalat Digital, runs the nationwide UAE Trade Connect platform against invoice fraud together with First Abu Dhabi Bank and Avanza Innovations. Each participating bank submits invoice attributes through its own network node, and an AI-driven engine checks them against all other submissions for duplicates without exposing customer data to competitors. The platform went live in April 2021 with seven commercial banks and screened invoices worth around AED 10 billion in its first year of operation.
Documentation status
- Project status: Production (as of 01 April 2021) — Launched nationwide in April 2021 with seven commercial banks; the Central Bank of the United Arab Emirates is represented on the steering committee.
- Evidence: Evidence medium
- Editorial review: pending
Description
The telecommunications group e& enterprise operates a nationwide platform together with First Abu Dhabi Bank and Avanza Innovations, through which banks in the United Arab Emirates detect invoice fraud. In invoice discounting, a company submits the same invoice to several banks, and because banks do not share customer data for competition and data protection reasons, this only becomes apparent after the damage has occurred. Through the shared platform, each bank only submits invoice features and receives a fraud assessment as a result, without seeing the business data of competitors.
The platform was announced in July 2020 and went live nationwide in April 2021 with seven commercial banks: Commercial Bank International, Commercial Bank of Dubai, Emirates NBD, First Abu Dhabi Bank, Mashreq Bank, National Bank of Fujairah and RAKBANK. In addition to these banks, the Central Bank of the United Arab Emirates is represented on the steering committee, giving the platform its nationwide character. Technically, each bank participates via its own node in an access-restricted ledger and submits invoice features there, but not the complete customer data. An AI-supported verification engine compares each submission against all others as well as external sources for duplicates and anomalies, returning only the verification result to the banks. In the first year of operation, invoices worth around ten billion dirhams, approximately 2.7 billion US dollars, were checked for double financing in this way. The case shows that competitors can jointly combat fraud if the architecture is designed to be data-minimising from the outset.
Perspectives
B2B — organisations perspective
For banks and other financiers, the model offers a solution to a problem that cannot be solved alone: fraud patterns only become visible through the collaboration of multiple institutions. Companies in other industries with comparable multiple submissions, such as in insurance or grant funding, can apply the data-minimising consortium model.
Employees perspective
In credit assessment and trade finance, an automated verification step replaces sampling and experiential knowledge. Fraud analysts focus more on suspicious patterns, and data protection functions must continuously ensure the data minimisation of the submitted features.
Benefits
General
- Double financing is detected before payment rather than after the damage has occurred.
- Banks retain their customer data because only verification features are submitted.
- In the first year of operation, invoices worth around ten billion dirhams were checked.
- The involvement of the Central Bank on the steering committee gives the platform nationwide binding authority.
B2B — organisations
- Reduced losses from invoice fraud in financing business.
- A transferable model for competition-neutral data exchange between competitors.
Employees
- Automated pre-verification instead of manual sampling.
- Clearer basis for decisions in credit assessment.
Challenges
General
- The benefit only arises when a relevant portion of the banks in a market participate.
- The balance between meaningful verification features and data minimisation must be continuously maintained.
- AI-supported verification generates false alarms that must be evaluated by analysts.
B2B — organisations
- Competing institutions must agree on joint governance and data standards.
Employees
- Data protection functions must continuously monitor which features are submitted.
Technology foundation
UAE Trade Connect combines an access-restricted ledger, in which each bank participates via its own node, with an AI-supported verification engine for matching the submitted invoice features. A distributed ledger is key here because competing banks want to jointly detect fraud without giving each other insight into their customer relationships.
Implementation examples
UAE Trade Connect as a nationwide verification network against invoice fraud
Companies submit the same invoice for discounting to multiple banks; since banks do not share customer data for competitive and data protection reasons, double financing remains undetected until damage occurs.
e& enterprise operates a nationwide platform together with First Abu Dhabi Bank and Avanza Innovations, where banks check invoices for double financing. Seven commercial banks participate via their own network nodes.
The platform was announced in July 2020 and went live in April 2021 with Commercial Bank International, Commercial Bank of Dubai, Emirates NBD, First Abu Dhabi Bank, Mashreq Bank, National Bank of Fujairah and RAKBANK. Each bank operates its own node to enter invoice features into an access-restricted ledger; an AI-powered verification engine compares these against all other submissions and external sources and returns only the verification result. In the first year of operation, invoices worth around ten billion dirhams, approximately 2.7 billion US dollars, were checked for duplicates.
Around ten billion dirhams, or approximately 2.7 billion US dollars, worth of invoices were checked for double financing in the first year of operation.
Lack of cross-bank visibility of submitted invoices, competitive concerns against data sharing between institutions, and protection of customer data during joint fraud detection.
Tags
Fraud prevention, Trade finance, Bank consortium, Data minimisation, Artificial intelligence