.jpg)
Catching points fraud without accusing the innocent
The operational problem
Loyalty points have cash value, which attracts fraud, including false claims for past flights. Wrongly accusing a genuine member does its own damage.
What the evidence reports
Microsoft said on 13 May 2020 that the SAS fraud model scores patterns across variables including login locations, email addresses, payment cards, travel timing, shared points pools and profile history. It is retrained automatically when code or data change. Microsoft reported 99% accuracy on retroactively registered flights and 300% more cases of small fraud found. SAS uses Microsoft's InterpretML toolkit to check why the model flags a case, and a person makes the final decision. Peter Gustavsson, senior systems specialist for EuroBonus, said: "We need to trust that using the model won't result in an accusation of fraud against an innocent customer." A second model predicts the quantity of fresh food to load onto flights, cutting food waste by up to 45%.
Operational significance
The fraud work shows the controls an airline needs when AI judges its own customers: an explanation for each flag, and a person who decides.
The figures are from a 2020 Microsoft customer story. SAS has not published later results, and the source does not say what the food model uses as inputs or how the 45% was measured.
Sources and verification (1)
Microsoft Customer Stories · Evidence grade B
A source can verify an announcement without verifying its projected outcome.