Autonomous personal-loan approval (Australia)
More
Recommended next steps
Highest-priority open actions
At a glance
- Owner
- Priya Chen
- Sponsor
- Head of Retail Banking
- Business unit
- Retail Bank
- Lifecycle stage
- Proposed
- Architecture
- No amplifiers declared
Top risks driving the decision
- R-AI-01 Autonomous consequential decision without human oversight
- R-CR-02 Potential discriminatory outcomes (fair lending)
- R-CY-01 Prompt injection via customer-supplied data
- R-RG-01 APRA CPS 230 operational risk mapping incomplete
Key conditions
- Human approval gate for loans above AUD 5,000
- Independent model validation report signed off by CRO
- APRA CPS 230 vendor registration complete
- Continuous model monitoring live in production
- Customer appeals and explainability path operational
What drives the score
GII points contributed by each CALIBRE factor (score 0 to 4 × weight)
Findings from each capability
- ARA Agentic Risk Analyst
- 18 findings
- CALIBRE Governance intensity
- 12 controls + GII summary
- Secure-RAG Evidence
- 0 citations
Decision narrative
Initiative SAMPLE-01 (Autonomous personal-loan approval) is assessed at ARA overall-residual High, CALIBRE band Critical (GII 81.25/100). The proposed level of autonomy on financially consequential decisions crosses the enterprise risk appetite and triggers material APRA CPS 230 and Privacy Act obligations that are not yet in scope of the proposal.
The Risk Analyst identified 2 High and 0 Critical risks. CALIBRE prescribes 10 governance controls at Critical band. Secure-RAG evidence coverage against cited regulations is 78%.
This should not be automatically approved. Escalate to the AI Risk Committee for judgement on whether reduced autonomy (Option B: AI-assisted with human approval) delivers the business objective at acceptable residual risk.
Description
An Australian retail bank wants to deploy an autonomous AI agent that approves personal loans up to AUD 50,000 using customer transaction history, credit bureau data and open banking data. No human review is planned for approvals under AUD 30,000. The bank intends to use a third-party foundation model via API. Decisions delivered to customers in-app within 30 seconds.
How action status is reported
A control is reported as implemented in the attestation only when its action is Complete. Add a link to the evidence where possible.
CALIBRE sets the required controls by band, not by reading the description. If a control does not apply, mark it Waived and put the written justification (or a link to it) in the evidence field. The waiver is shown in the attestation.
| Priority | Action | Status |
|---|---|---|
| Critical | Prepare Risk Committee briefing pack from this Decision Pack SAMPLE-01-A-01
|
Open |
| High | Implement Human approval gate for consequential decisions SAMPLE-01-A-02
|
Open |
| High | Complete Independent model validation SAMPLE-01-A-03
|
Open |
| High | Complete APRA CPS 230 vendor registration SAMPLE-01-A-04
|
Open |
| High | Design specific control for R-AI-01: Autonomous consequential decision SAMPLE-01-A-05
|
Open |
| High | Design specific control for R-CY-01: Prompt injection via customer-supplied data SAMPLE-01-A-06
|
Open |
| Medium | Index authoritative source for ASIC RG 271 into Secure-RAG SAMPLE-01-A-07
|
Open |
No actions match
Change the filters to see more.
ARA R-AI-01Autonomous consequential decision without human oversight
AI approves loans up to AUD 50k with no human review. Model errors are consequential and irreversible for the customer.
ARA R-CR-02Potential discriminatory outcomes (fair lending)
Historical bias in training data may produce disparate outcomes across protected attributes.
ARA R-CY-01Prompt injection via customer-supplied data
Customer-supplied text fields in the application may contain adversarial payloads.
ARA R-RG-01APRA CPS 230 operational risk mapping incomplete
Third-party model dependency crosses CPS 230 material service provider thresholds.
ARA R-OP-03Third-party foundation-model dependency
Foundation-model vendor may deprecate model, change pricing, or fail SLA.
ARA CMP-01APRA CPS 230 (Australia) - applicable
This proposal creates a material service dependency on a third-party AI vendor and automates a consequential decision.
ARA CMP-02APRA CPS 234 (Australia) - applicable
Customer data leaves the bank's information-security perimeter to the model provider.
ARA CMP-03Privacy Act 1988 (APPs) (Australia) - applicable
Personal information is used for automated decision-making. APP 6 and APP 11 apply.
ARA CMP-04ASIC RG 271 (Australia) - potentially applicable
Consumers rejected by the AI must have accessible internal-dispute-resolution channels.
ARA T-AI-01Prompt Injection (direct)
Customer-supplied text may contain adversarial instructions overriding the system prompt.
ARA T-AI-02Sensitive Data Exfiltration
Model may echo customer PII in generated explanations, potentially leaking to logs or downstream.
ARA T-TRAD-01Privileged access to model pipeline
Model training and inference pipeline requires strong access control.
ARA C-01Human approval gate for consequential decisions
All loan decisions above AUD 5,000 require human-in-the-loop approval before customer notification.
ARA C-02Independent model validation
Third-party model validation report covering fairness, robustness, and performance drift before production.
ARA C-03Output validation and PII scanning
Automated scanning of model outputs for PII and adversarial-content indicators before delivery to customer.
ARA C-04APRA CPS 230 vendor registration
Register foundation-model provider as material service provider and complete operational risk mapping.
ARA C-05Model monitoring and drift detection
Continuous monitoring for accuracy drift, disparate outcomes and rejection-rate anomalies.
ARA C-06Customer explainability and appeals path
Every automated rejection provides a plain-English explanation and a link to human review.
CALIBRE CAL-GII-CriticalGovernance Intensity Index = 81.25/100 (band: Critical)
CALIBRE band: Critical. Rule 1: autonomous action on sensitive data, floor High: ensure human oversight (HITL-01).
CALIBRE CAL-GOV-03GOV-03 - Executive and board risk sign-off
CALIBRE core control, required from band Critical upward. Maps to NIST GOVERN-2.3 | ISO 42001 5.1; 6.1.3 | EU AI Act Art.17.
CALIBRE CAL-RES-01RES-01 - Redress and incident response
CALIBRE core control, required from band Critical upward. Maps to NIST MANAGE-4.3; MEASURE-3.3 | ISO 42001 A.8.3; A.8.4 | EU AI Act Art.73.
CALIBRE CAL-XAI-01XAI-01 - Explainability and transparency
CALIBRE core control, required from band High upward. Maps to NIST MEASURE-2.9 | ISO 42001 A.8.2 | EU AI Act Art.13; Art.50; Art.86.
CALIBRE CAL-HITL-01HITL-01 - Human oversight (human-in-the-loop)
CALIBRE core control, required from band High upward. Maps to NIST MAP-3.5 | ISO 42001 A.9.3 | EU AI Act Art.14; Art.26.
CALIBRE CAL-MON-01MON-01 - Runtime monitoring and drift detection
CALIBRE core control, required from band High upward. Maps to NIST MEASURE-2.4; MANAGE-4.1 | ISO 42001 A.6.2.6; A.6.2.8 | EU AI Act Art.72; Art.26.
CALIBRE CAL-ASR-01ASR-01 - Independent pre-deployment assurance
CALIBRE core control, required from band High upward. Maps to NIST MEASURE-1.3 | ISO 42001 A.6.2.4; A.6.2.5 | EU AI Act Art.17; Art.43.
CALIBRE CAL-GOV-02GOV-02 - AI impact and risk assessment
CALIBRE core control, required from band Moderate upward. Maps to NIST MAP-5.1 | ISO 42001 A.5.2; A.5.4 | EU AI Act Art.9; Art.27.
CALIBRE CAL-DATA-01DATA-01 - Data governance and PII protection
CALIBRE core control, required from band Moderate upward. Maps to NIST MEASURE-2.10 | ISO 42001 A.7.2; A.7.3; A.7.4; A.7.5 | EU AI Act Art.10.
CALIBRE CAL-SEC-01SEC-01 - AI security controls
CALIBRE core control, required from band Moderate upward. Maps to NIST MEASURE-2.7 | ISO 42001 A.6.2.3 | EU AI Act Art.15.
CALIBRE CAL-SUP-01SUP-01 - AI supply-chain and third-party assurance
CALIBRE core control, required from band Moderate upward. Maps to NIST GOVERN-6.1; MANAGE-3.1 | ISO 42001 A.10.2; A.10.3 | EU AI Act Art.25.
CALIBRE CAL-LIT-01LIT-01 - AI literacy and competence
CALIBRE core control, required from band Low upward. Maps to NIST GOVERN-2.2 | ISO 42001 7.2 | EU AI Act Art.4.
CALIBRE CAL-GOV-01GOV-01 - AI use-case registration
CALIBRE core control, required from band Low upward. Maps to NIST GOVERN-1.6 | ISO 42001 A.4.2 | EU AI Act Art.49.
No findings match
Change the filters to see more.
Runtime: operating envelope, control tests and authority checks
Operating envelope
No operating envelope
Push one from ARA. Until then every agent action is escalated.
Recent authority checks
Open authorityNo checks recorded for this initiative.
Agent gateways call POST /api/authority/check before acting. Secure-RAG reports control tests to POST /api/test-results.