Human-centred use
Customer-facing AI should identify itself, provide an easy path to a human and require human approval for sensitive or consequential actions.
Trust & transparency
AI & Data Use Policy
How AI-assisted features should be identified, controlled and reviewed.
Data safeguards
Apply data minimisation, access controls, retention rules and vendor due diligence. Client or customer personal information must not be used to train models without contractual authorisation and legal review.