Misalignment refers to AI systems producing outputs that deviate from financial services regulatory requirements, industry standards or firm policies, even when technically functional. This can occur when general-purpose models are applied to regulated contexts without appropriate adaptation. Misalignment might result in advice documentation that omits required disclosures, compliance checks that miss sector-specific risks or language that conflicts with treating customers fairly principles. Financial services-specific training, careful prompting and validation processes help ensure AI outputs remain appropriately aligned.
