Vyntra's AI risk models may have stopped a USD 500,000 ATM fraud

AI system

System: Vyntra (NetGuardians) AI risk models for real-time payment and e-banking fraud detection
Model: Not assessed
Release: Not assessed
Application developer: Vyntra (NetGuardians SA, merged with Intix in June 2025)
Role: Not assessed
Configuration: Not assessed
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: Behavioral profiles of each customer's and the bank's usual transaction and e-banking session variables (beneficiary, country, amount, currency, browser, screen resolution, language). Real-time scoring before a payment is released.
Human oversight: Not assessed
Relation to AI: The evidence is limited to the maker's account. Vyntra's page attributes each block to its AI risk models detecting variables outside the customer's or bank's usual profile. No customer, bank, date or official record is available to assess whether the models, rather than staff or other controls, produced the outcome. The AI's contribution is not established.
AI contribution: Not assessed

Outcome identifiers: Not assessed
Benefit: fraudsters were arrested while attempting to withdraw USD 500,000 (KSh 58 million)
Category: Reduced Risk
Status: Not assessed
Beneficiary groups: People. Organizations. Society.
Beneficiaries: Not assessed
Reach: Not assessed
Distribution and access: Not assessed
Duration: Not assessed
Expected?: Yes
Costs and harms: Not assessed

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: fraudsters were arrested while attempting to withdraw USD 500,000 (KSh 58 million)
Comparison: Not assessed
Results: fraudsters were arrested while attempting to withdraw USD 500,000 (KSh 58 million)
Additional value from AI: Not assessed
Limitations: Not assessed
Denominators: Not assessed
Materials for replication: Not assessed

Kind of value: fraudulent ATM withdrawal detected and stopped, with arrests
Amount: USD 500,000 (KSh 58 million)
Currency and prices: Not assessed
Period and unit: Not assessed
Realization: Not assessed
Recipients and payers: the second bank, an unnamed client bank (the page labels the case 'Kenya')
Gross or net: Not assessed
Included costs: Not assessed
Comparison: Not assessed
Method: Not assessed
Return and payback: Not assessed

Participants: Developer: Vyntra (NetGuardians SA, merged with Intix in June 2025). Operator: Unnamed client banks.
Jurisdictions: Benefit: Not stated; amounts are given in Swiss francs, euros, pounds, US dollars and Kenyan shillings. Developer: Switzerland.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: The fraudulent transaction departs from the profile on several variables; the vendor notes cases with little customer history relied on bank-level anomalies. The bank acts on the alert before the money leaves.
Replication attempts: Not assessed
Later findings: No dated follow-up to any of the cases was found; the June 2025 merger of NetGuardians and Intix into Vyntra concerns the vendor, not the cases.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Assessed benefits: People. Organizations. Society.
Depth: People: 3. Organizations: 4. Society: Not assessed.
Reasons: People: The vendor describes customers whose social-engineering transfers of EUR 125,000 and USD 170,000 were blocked, a loss averted for each. No customer is named or independently documented. Organizations: The vendor states that a USD 500,000 (KSh 58 million) ATM fraud was stopped with arrests. It says payments up to USD 850,000 were blocked before funds left client banks. Serious losses averted if the account holds. Customers, banks and dates are not given. Society: No aggregate verified by a regulator is given.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: Independently reported

Kind: Not assessed
Related entries: Not assessed
Field: Banking fraud prevention. Retail and corporate payment fraud at client banks.
Purpose: The risk models are built to score payments and e-banking sessions against customer and bank profiles in real time and block fraudulent transfers. Each case is that intended use.
Dates: Date basis unassessed: 2022 (precision unassessed). Updated: 21 October 2026.
Stage: Not assessed
Version: SHA-256: 0242cb63702b…