Institute for Operational Assurance

Framework

Human and AI Assurance

Humans and AI agents should operate against the same underlying operational truth.

Intent

Shared operational truth

Policies · controls · permissions · escalation

Human actors

AI agents

Evidence and outcomes return to the assurance loop
Organisational intent and shared operational truth guide human actors and AI agents through common policies, controls, permissions and escalation paths, with evidence returning into the assurance loop.

One operational system

Policies, controls, responsibilities and strategic intent should govern human and automated activity.

Appropriate assurance

Interfaces and autonomy may differ, but roles, permissions, escalation and evidence should remain explicit.

Successful is not necessarily assured

Task completion alone is insufficient when activity falls outside organisational intent.

Begin with one workflow

As machines gain agency

Join the Institute’s flagship research question.

Explore Who Assures the Agents?