Turning a Declined Payment into a Lesson

The Jackal is Mastercard's name for today's adaptive and persuasive threat actor targeting human trust rather than technical gaps, giving rise to social engineering, business email compromise, AI-enabled scams.

For the 2026 Global Intern Innovation Challenge, Mastercard asked for a marketable fraud education and awareness product that could counter the Jackal's schemes. Interns were grouped into remote teams of 11 from across the world and given 6 hours to complete the challenge.

Mastercard already scores every transaction through Decision Intelligence (DI), which rates approvability on a wide scale. Issuers who have DI apply their own rules on top, determining the score thresholds and additional signals at which transactions will be approved, declined, or flagged. In this process the consumer remains uninformed about declines and risky transactions attempted on their card.

• A coached victim reads a decline as an obstacle, not as a warning.

Even after an issuer decides a transaction is risky enough to decline, consumers are left out and no more protected against the scheme than they started. There's no clear understanding of next steps at the critical moment where users would care the most.

Fraud IQ is DI's consumer-intelligence layer on a score the issuer already computes. It doesn't influence DI's fraud scoring or a bank's decision to approve a transaction. When DI flags a payment, Fraud IQ tells the cardholder why, in the issuer's own app, at the moment of the decline: the signals that fired, what each one means, and what to watch for next time. Then it asks whether it was them, and sends the answer back to the issuer.

01

An explanation API

Takes a DI outcome and its contributing signals and returns cardholder-facing content: a plain-English reason, a category of what to watch for, a “was this you?” prompt.

02

A feedback API

Captures the cardholder's response and routes it back to the issuer's fraud ops and into Mastercard's models, the same way DI feedback loops work today.

03

An add-on to Decision Intelligence

Issuers who already opt into DI can add this layer without a new integration with the merchant or a new step in authorisation.

The coached payment

Someone is being talked into $500 of gift cards. They are awake, mid-checkout, and have already been told to expect a problem. The demo starts inside the merchant's payment sheet, and the decline interrupts it.

The stolen number

A criminal tests a card with a $1.02 charge at 3:12 AM, then spends it three minutes later. The cardholder is asleep and still holding the card. The demo starts on a lock screen.

Both end on the same review screen.

The obvious design for the “yes, this was me” action might retry the payment, but a coached victim will answer yes, leading to the moment the product is most needed and least trusted. Instead of retrying, it asks the one question that separates a real purchase from a coached one: did somebody ask you to buy these? This additional step also allows banks to measure true engagement.

01

Fraud IQ only reaches cardholders whose issuer chooses to build it, and Mastercard has limited control over adoption and the UI.

02

The explanation layer also has to stay general enough not to teach fraudsters how the model works, which caps how specific or satisfying the “why” can ever be for a legitimate cardholder.

03

A coached victim mid-scam may not trust or even read an in-app explanation no matter how well it's written, so the product reduces risk rather than eliminating it.

For issuers, educating cardholders can lead to fewer fraud losses, fewer false-decline complaints, and lower attrition.

Cybersecurity and AI-driven fraud prevention have been a stated growth area for Mastercard, alongside acquisitions like RiskRecon and Ekata and products like Decision Intelligence itself. Fraud IQ would earn revenue the same way DI does, as a paid add-on issuers purchase.

Fraud IQ's key KPI would be fraud loss reduction on flagged transactions, which directly measures whether it is helping combat fraud. Secondary KPIs: issuer adoption rate, and the “was this you?” response rate as a proxy for whether cardholders are actually reading it.

• Top 6 finalist
• Working prototype and slide deck delivered
• Ships on a product issuers have already bought