Workforce-Impact Assessment & Human-Role Preservation

Control objective

Workforce-Impact Assessment & Human-Role Preservation checks that, before an agentic AI system is scaled into production, the organisation has formally assessed how it will affect human jobs and decision-making, and has explicitly preserved meaningful human roles over consequential decisions. This matters because agentic systems that automate document extraction, retrieval, and chat workflows in a bank can quietly displace or de-skill staff and erode human oversight of decisions that should remain human-owned (such as final credit, fraud, or customer-treatment calls). There is no numeric metric or formula for this control; it is governance-based and evidenced by documentation. To implement it, conduct and sign off a workforce-impact assessment as part of the pre-scale gate, and maintain a human decision-rights register that records which decisions must retain a human in or on the loop, who holds them, and how that authority is exercised. The threshold is that the workforce-impact assessment is completed pre-scale and the human decision-rights register is maintained and kept current; absent or stale documentation constitutes a breach. A breach should trigger holding the scale-up until the assessment is completed and decision rights are properly recorded and approved.