Model risk management
SR 11-7 and OCC 2011-12
Models that drive decisions, including vendor and AI models, must be validated, monitored and governed.
Airline held liable for a customer-service chatbot that gave wrong bereavement fare rules
VerifiedChatbot misinformation2024
Australian welfare agency raised debts by automatically averaging annual income across fortnights
VerifiedDecision unreliability2023
Allegation that the Apple Card credit algorithm gave women lower limits, reviewed by New York regulator
InsufficientBias in credit decisions2021
Widely used health risk algorithm scored Black patients as healthier than equally sick White patients
VerifiedBias in risk scoring2019
A case bears on an obligation when it informs how that obligation applies to AI. This is not legal advice and not a finding that anyone violated it.