# Agentic Risk Analyst - Executive Assessment ## 1. Scenario > 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 will be delivered to customers in-app within 30 seconds. **Interpretation:** An Australian bank wants an AI agent, built on a third-party foundation model, to approve personal loans up to AUD 50,000 in about 30 seconds, with no human review below AUD 30,000. - Business objective: Approve personal loans faster with less manual effort - Industry: Banking (Australian authorised deposit-taking institution) - Use case: Autonomous approval of personal loans up to AUD 50,000 - AI autonomy: fully_autonomous - Human oversight: None below AUD 30,000; not stated for AUD 30,000 to 50,000 - Decision impact: Major: credit decisions affect consumers' finances and access to credit - Jurisdictions: Australia - Data types: Transaction history, Credit bureau data, Open banking (CDR) data, Identity data ## 2. Executive Decision **Decision:** APPROVE WITH CONDITIONS **Overall inherent risk:** High **Overall residual risk:** Medium Proceed only with a human gate for declines and larger loans, validated fairness and affordability, and updated transparency. Full automation of declines is not supported. ### Top risks - Unfair or discriminatory credit outcomes - Unsuitable loans approved - Automated decisions without required transparency ### Key conditions - Human review of all declines and approvals above AUD 30,000 - Fairness and affordability testing before go-live - Privacy policy updated for APP 1.7 to 1.9 before 10 December 2026 ## 3. Remediation Plan ### 0-30 days - **A-01** (Critical, Head of Credit): Design the human review step and reason codes ### 30-90 days - **A-02** (High, Model Risk): Run fairness and affordability validation - **A-03** (High, Privacy Officer): Update privacy policy and notices ### 90+ days - **A-04** (Medium, Model Risk): Set up monthly drift and fairness monitoring ## 4. Risk Register | ID | Category | Title | Likelihood | Impact | Inherent | |---|---|---|---|---|---| | R-01 | Conduct | Unfair or discriminatory credit outcomes | Possible | Major | High | | R-02 | Credit | Unsuitable loans approved | Possible | Major | High | | R-03 | Privacy | Automated decisions without required transparency | Likely | Moderate | High | | R-04 | Third-party | Foundation model change or outage | Possible | Moderate | Medium | | R-05 | Model | Drift after deployment | Likely | Moderate | High | ## 5. Regulatory Findings ### National Consumer Credit Protection Act 2009 (responsible lending) (Australia) - Applicable **Why relevant:** Personal loans to consumers are regulated credit. **Scenario impact:** Automation must still meet responsible-lending obligations. **Obligations:** - Make reasonable inquiries and verify the consumer's financial situation - Assess that the loan is not unsuitable **Required controls:** - Affordability rules - Suitability assessment record ### Privacy Act 1988 (APPs, including APP 1.7 to 1.9) (Australia) - Applicable Source: **Why relevant:** Personal and credit information drives automated decisions. **Scenario impact:** Policy and notice updates; data minimisation. **Obligations:** - Describe automated decisions in the privacy policy (from 10 December 2026) - Handle credit information under the credit reporting rules **Required controls:** - Privacy impact assessment - Updated privacy policy ### APRA CPS 230 Operational Risk Management (Australia) - Applicable Source: **Why relevant:** The bank is APRA-regulated and relies on a third-party model provider. **Scenario impact:** The model provider may be a material service provider. **Obligations:** - Manage material service providers under a formal agreement - Maintain critical operations within tolerance **Required controls:** - Service provider due diligence - Business continuity plan ### Anti-discrimination legislation (Commonwealth and state) (Australia) - Potentially Applicable **Why relevant:** Credit decisions must not discriminate on protected attributes. **Scenario impact:** Fairness testing is needed. **Obligations:** - Avoid direct and indirect discrimination **Required controls:** - Fairness testing across groups ## 6. Compliance Gaps | Area | Gap | Risk | Remediation | |---|---|---|---| | Transparency | Applicants may not receive reasons. | High | Generate reason codes and review declines. | ## 7. Security Threats ### T-01 - Adversarial manipulation of application data (ai_specific) - High Applicants game features to gain approval. - **Attack vector:** Crafted transaction patterns or documents - **Assets at risk:** Credit decisions - **Likelihood x Impact:** Possible x Major - **MITRE:** MITRE ATLAS AML.T0043 ### T-02 - Credential compromise of the model API (traditional) - Medium Attacker uses the bank's API key. - **Attack vector:** Leaked secrets - **Assets at risk:** Model access, Cost - **Likelihood x Impact:** Unlikely x Major - **MITRE:** MITRE ATT&CK T1552 ## 8. Controls | ID | Control | Type | Effectiveness | Priority | Addresses | |---|---|---|---|---|---| | C-01 | Human review of declines and loans above AUD 30,000 | Preventive | High | Immediate | R-01, R-02 | | C-02 | Fairness and affordability testing before release and monthly | Detective | Medium | Immediate | R-01, R-02, R-05 | | C-03 | Automated-decision transparency | Preventive | High | Near-term | R-03 | | C-04 | Model change control and fallback | Preventive | Medium | Near-term | R-04, R-05 | ## 9. Residual Risk | Risk | Inherent | Controls | Residual | Justification | |---|---|---|---|---| | R-01 - Unfair or discriminatory credit outcomes | High | C-01, C-02 | Medium | Human review and testing catch most unfair outcomes. | | R-02 - Unsuitable loans approved | High | C-01, C-02 | Medium | Affordability testing and review of large loans. | | R-03 - Automated decisions without required transparency | High | C-03 | Low | Policy and reasons address the obligation. | | R-04 - Foundation model change or outage | Medium | C-04 | Low | Version pinning and fallback. | | R-05 - Drift after deployment | High | C-02, C-04 | Medium | Monthly monitoring. | ## Evidence status No external evidence was retrieved. Regulatory findings come from the model's general knowledge and must be verified against the source texts. --- _Generated by Agentic Risk Analyst (ARA). Decision support only, not legal or regulatory advice. Every rating and citation must be reviewed by a qualified person before it is relied on._