What was claimed, and what the evidence says.
All automated credit decisions are reviewed by a human before final approval.
Source: Public AI policy, §4.2
Customer chat assistant discloses that it is an AI system at the start of each session.
Source: Product documentation
Model outputs are logged and retained for 24 months.
Source: Trust center statement
Training data excludes customer-identifiable information.
Source: Investor presentation, p. 17
Fraud-detection model is re-evaluated against a fixed benchmark every quarter.
Source: Annual report, risk section
Users can request human escalation at any point in an automated flow.
Source: Help center
Third-party models are assessed before production use.
Source: Procurement policy
Automated decisions can be explained to affected customers on request.
Source: Customer terms, §9