How many AI incidents at banks since 2022 were cases where the bank's or fintech's own AI system failed, misbehaved, caused an outage, or was withdrawn or reversed?
As of Oct 21, 2026, the register holds 9 publicly reported AI incidents at banks and fintechs where the bank's or fintech's own AI system failed, misbehaved, caused an outage, or was withdrawn or reversed since January 2022. 5 of the 9 records involve systems the institution or a regulator described as AI; 4 are automated systems not described as AI. Countries: United States (4), Australia (1), United Kingdom (1). Average severity 2.7 on a 1–5 scale. Status: 1 reported, 2 confirmed by the institution, 1 regulator finding, 1 court filing, 3 settled, 1 resolved. The most recent is "Libra Internet Bank's OpenAI-based assistant 'Lia' calls a customer an 'idiot'" (Sep 17, 2026).
Incidents per category within each AI role · bars link to the category page · computed from the register