# Agentic Risk Analyst - Executive Assessment ## 1. Scenario > A Tier-2 bank in Australia proposes to replace its rules-based AML transaction-monitoring engine with an LLM-based agent that reads free-text payment memos, customer notes and adverse-media results to score suspicious-matter reports (SMRs). The agent will draft the SMR narrative and auto-lodge to AUSTRAC once a compliance analyst clicks approve. The model is hosted on a US cloud region. **Interpretation:** A Tier-2 Australian bank wants an LLM agent to replace rules-based transaction monitoring, score alerts and draft SMRs, lodged after analyst approval; the model is hosted in the US. - Business objective: Improve detection of suspicious matters and reduce analyst effort - Industry: Banking (Tier-2, Australia) - Use case: LLM agent scores alerts and drafts suspicious matter report (SMR) narratives - AI autonomy: human_approval_required - Human oversight: A compliance analyst approves before lodgement - Decision impact: Major: missed or wrong reports have legal consequences - Jurisdictions: Australia, United States (hosting) - Data types: Payment memos, Customer notes, Adverse media results, Transaction data ## 2. Executive Decision **Decision:** REQUIRES REMEDIATION **Overall inherent risk:** Critical **Overall residual risk:** Medium The idea has merit as a prioritisation and drafting aid, but replacing the rules engine without proven parity is not acceptable. ### Top risks - Suspicious matters missed after replacing rules - Analysts approve drafts without real review ### Key conditions - Do not retire the rules engine until detection parity is demonstrated - Evidence-linked drafts and second-line sampling ## 3. Remediation Plan ### 0-30 days - **A-01** (Critical, AML Compliance Lead): Redesign as parallel run alongside the rules engine ### 30-90 days - **A-02** (High, AI Engineering): Build citation checks into SMR drafts - **A-03** (High, Privacy Officer): Assess cross-border data handling of the US-hosted model ### 90+ days _No actions._ ## 4. Risk Register | ID | Category | Title | Likelihood | Impact | Inherent | |---|---|---|---|---|---| | R-01 | Compliance | Suspicious matters missed after replacing rules | Possible | Severe | Critical | | R-02 | Operational | Analysts approve drafts without real review | Likely | Major | High | | R-03 | Privacy | Customer data processed offshore | Likely | Moderate | High | | R-04 | Model | Hallucinated facts in SMR narratives | Possible | Major | High | ## 5. Regulatory Findings ### Anti-Money Laundering and Counter-Terrorism Financing Act 2006 (suspicious matter reporting) (Australia) - Applicable Source: **Why relevant:** The agent supports SMR decisions and drafting. **Scenario impact:** The program must cover the new monitoring method. **Obligations:** - Report suspicious matters to AUSTRAC within the statutory time frames - Do not disclose ('tip off') that a report has been made - Maintain an AML/CTF program **Required controls:** - Parallel run before retiring rules - Tipping-off controls on outputs ### Privacy Act 1988 (APP 8 cross-border disclosure) (Australia) - Applicable Source: **Why relevant:** Personal information is sent to a US-hosted model. **Scenario impact:** Contract and assessment of the US provider. **Obligations:** - Take reasonable steps so the overseas recipient handles the information consistently with the APPs **Required controls:** - Data processing agreement - Data minimisation ### APRA CPS 230 Operational Risk Management (Australia) - Applicable Source: **Why relevant:** Material reliance on a third-party model provider for a compliance-critical process. **Scenario impact:** Provider assessment and exit plan. **Obligations:** - Formal agreement with the material service provider **Required controls:** - Service provider due diligence ## 6. Compliance Gaps | Area | Gap | Risk | Remediation | |---|---|---|---| | Validation | No evidence the agent detects what the rules detect. | Critical | Run both in parallel and compare before any retirement. | ## 7. Security Threats ### T-01 - Prompt injection through payment memos (ai_specific) - High A memo instructs the agent to score the transaction as benign. - **Attack vector:** Customer-controlled free text - **Assets at risk:** Alert scores, SMR drafts - **Likelihood x Impact:** Possible x Major - **MITRE:** OWASP LLM01:2025 ### T-02 - Exposure of SMR content (traditional) - High Confidential report content leaks. - **Attack vector:** Logs or provider retention - **Assets at risk:** SMR data - **Likelihood x Impact:** Unlikely x Severe ## 8. Controls | ID | Control | Type | Effectiveness | Priority | Addresses | |---|---|---|---|---|---| | C-01 | Parallel run with the rules engine | Preventive | High | Immediate | R-01 | | C-02 | Evidence-linked SMR drafts | Preventive | Medium | Immediate | R-04, R-02 | | C-03 | Quality sampling of approvals | Detective | Medium | Near-term | R-02 | | C-04 | Data minimisation and onshore option | Preventive | Medium | Near-term | R-03 | ## 9. Residual Risk | Risk | Inherent | Controls | Residual | Justification | |---|---|---|---|---| | R-01 - Suspicious matters missed after replacing rules | Critical | C-01 | Medium | Parallel running prevents coverage loss. | | R-02 - Analysts approve drafts without real review | High | C-02, C-03 | Medium | Citations and sampling expose weak review. | | R-03 - Customer data processed offshore | High | C-04 | Medium | Minimisation reduces exposure. | | R-04 - Hallucinated facts in SMR narratives | High | C-02 | Low | Uncited statements are flagged. | ## 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._