Mockup notes (ADM). This shows what an AI Audit Leaflet looks like for automated decision-making systems assessed across Pre/In/Post lifecycle stages — design preview, not real audit data. The leaflet is derived from the audit report. Grades come from metric-based scores, evidence-based scores, and judgment-based scores. All feed into aggregation rules to produce stage grades and dimension grades. Stage sub-grades and trajectory charts show how risk evolves through the pipeline.

AI Audit Leaflet

Independent AI Assessment
System: System name
Version: v1.0
Type: ADM
Domain: e.g. Education, Healthcare
Owner: Organization name
Risk level: High / Limited / Low
Audit scope
✓
Pre-Processing
Baseline
✓
In-Processing
Baseline
✓
Post-Processing
Standard
Assessment results
Bias & Fairness C
B Training data
C Model evaluation
D Production
"Training data shows reasonable demographic representation, but fairness degraded through the pipeline. Production outcomes show significant group disparities that were not present in the data and were not caught during evaluation."
Reliability B
C Training data
A Model evaluation
B Production
"Training data had quality gaps, but the model corrected for them. Slight performance drop in production compared to evaluation — minor drift detected."
Privacy & Confidentiality B
B Training data
B Model evaluation
A Production
"Training data anonymisation is adequate. Access controls and retention policies are in place. No evidence of re-identification risk in production."
Security & Misuse B
— Training data
B Model evaluation
B Production
"Access controls are in place. Adversarial robustness testing not yet conducted. Data poisoning risk is low given controlled data sources."
Governance C
C Training data
C Model evaluation
B Production
"Model documentation and decision logs are maintained. Change management process exists but version control of training data needs improvement."
Key metrics across lifecycle
Demographic Parity
Fairness
good ok poor .92 .85 .78 Pre In Post
Accuracy
Reliability
good ok poor .82 .94 .88 Pre In Post

Audit details

Date: DD/MM/YYYY
Auditor: Auditor name
Valid until: DD/MM/YYYY

Client

Organization: Client name
Contact: Contact person

How to read this leaflet. Each risk dimension is graded A (best) to E (worst) based on an independent audit. Sub-grades show performance at each lifecycle stage: training data (Pre), model evaluation (In), and production (Post). A dash (—) indicates the stage was not assessed. The trajectory charts show how a key metric for fairness and reliability evolved through the pipeline — whether issues were introduced, corrected, or amplified at each stage. Grades are derived mechanically from individual audit checks. The full technical report with detailed findings is available from the auditor.

Grade scale: A No significant issues · B Minor issues · C Moderate issues or limited scope · D Critical issues · E Systemic failure