University of Sydney Responsible AI Assessment

Currently in pilot, subject to change. Version date: 7 May 2026.
Untitled AI initiative
Assessment Details
Part A — Initiative Profile
These questions capture the profile of the AI initiative. They are not scored but inform which governance questions apply in the detailed assessment.
System origin

What is the system origin?

Lifecycle stage

What stage is the AI initiative at?

AI technique

What is the AI technique or type?

Use case

What is the primary use case?

User cohort

Who is the primary user cohort?

Part B — Inherent Risk Assessment
These questions determine the inherent risk score and tier. If multiple options apply, select the option that represents the highest inherent risk — options are ordered from lowest to highest risk.
Domain sensitivity

What is the primary domain or field of application of this AI system?

Affected cohort

Who is primarily affected by the system's outputs or decisions?

Harm severity

What is the worst credible harm to people or rights if the system fails or produces an incorrect output?

Reversibility

If the system produces an incorrect or harmful output, how easily can it be identified and corrected?

Scale of reach

How many people could be affected if the system behaves incorrectly or fails?

Data sensitivity

What is the most sensitive type of data the system handles?

Differential impact

Could errors or weaker performance from this system affect some groups of people more than others?

Autonomy level

To what extent does the system operate without human intervention?

Human oversight

What level of human oversight is applied to the system's outputs or decisions?

Explainability

How explainable are the system's outputs or the basis for its decisions?

The University of Sydney AI Assessment Framework · Currently in pilot, subject to change · Version date: 7 May 2026