AI Risk Assessment Template
Structured risk assessment for AI agent deployment decisions.
Overview
Use this template to assess risks before deploying new agents or promoting existing agents to higher governance zones. Risk assessment is required for Zone 2 and mandatory with formal documentation for Zone 3.
Assessment Information
| Field | Value |
|---|---|
| Agent Name | |
| Agent ID | |
| Proposed Zone | Zone 1 / Zone 2 / Zone 3 |
| Business Owner | |
| Assessment Date | |
| Assessor |
Agent Description
Purpose
Describe what the agent does and the business problem it solves.
Users
Who will use this agent? How many users?
Data Sources
What data does the agent access? List all knowledge sources and connectors.
Outputs
What outputs does the agent produce? How are they used?
Risk Assessment Categories
1. Data Risk
| Risk Factor | Rating | Justification |
|---|---|---|
| Data sensitivity | Low / Medium / High | |
| Data volume | Low / Medium / High | |
| External data exposure | Yes / No | |
| Customer PII accessed | Yes / No | |
| Financial data accessed | Yes / No |
Overall Data Risk: Low / Medium / High
Mitigating Controls:
- DLP policies applied (Control 1.5)
- Sensitivity labels enforced
- Data minimization implemented (Control 1.14)
- Other: __
2. Regulatory Risk
| Risk Factor | Applies? | Justification |
|---|---|---|
| FINRA supervision required | Yes / No | |
| SEC records requirements | Yes / No | |
| GLBA customer data | Yes / No | |
| SOX financial data | Yes / No | |
| OCC model risk guidance | Yes / No | |
| Fair lending implications | Yes / No |
Overall Regulatory Risk: Low / Medium / High
Mitigating Controls:
- Supervisory procedures documented (Control 2.12)
- Audit logging configured (Control 1.7)
- Model risk assessment completed (Control 2.6)
- Bias testing completed (Control 2.11)
- Other: __
3. Operational Risk
| Risk Factor | Rating | Justification |
|---|---|---|
| Business impact if unavailable | Low / Medium / High | |
| Dependency on agent for critical processes | Low / Medium / High | |
| Complexity of agent logic | Low / Medium / High | |
| Integration points | Low / Medium / High |
Overall Operational Risk: Low / Medium / High
Mitigating Controls:
- Business continuity plan (Control 2.4)
- Change management process (Control 2.3)
- Testing procedures (Control 2.5)
- Monitoring configured (Control 3.2)
- Other: __
4. Security Risk
| Risk Factor | Rating | Justification |
|---|---|---|
| Attack surface | Low / Medium / High | |
| Privilege level required | Low / Medium / High | |
| External exposure | None / Internal / External | |
| Data exfiltration potential | Low / Medium / High |
Overall Security Risk: Low / Medium / High
Mitigating Controls:
- Managed Environment enabled (Control 2.1)
- Conditional Access applied (Control 1.11)
- Runtime protection enabled (Control 1.8)
- Network isolation configured (Control 1.20)
- Other: __
5. Reputational Risk
| Risk Factor | Rating | Justification |
|---|---|---|
| Customer-facing | Yes / No | |
| Brand impact if fails | Low / Medium / High | |
| Public visibility | Low / Medium / High | |
| Hallucination risk | Low / Medium / High |
Overall Reputational Risk: Low / Medium / High
Mitigating Controls:
- Human-in-the-loop for high-risk decisions
- Hallucination monitoring (Control 3.10)
- AI disclosure implemented (Control 2.19)
- Escalation procedures documented
- Other: __
Overall Risk Summary
| Category | Rating | Weight | Weighted Score |
|---|---|---|---|
| Data Risk | Low=1 / Medium=2 / High=3 | 25% | |
| Regulatory Risk | Low=1 / Medium=2 / High=3 | 25% | |
| Operational Risk | Low=1 / Medium=2 / High=3 | 20% | |
| Security Risk | Low=1 / Medium=2 / High=3 | 20% | |
| Reputational Risk | Low=1 / Medium=2 / High=3 | 10% | |
| Total | 100% |
Risk Classification:
- 1.0-1.5: Low Risk — Zone 1 appropriate
- 1.6-2.2: Medium Risk — Zone 2 appropriate
- 2.3-3.0: High Risk — Zone 3 required
Calculated Risk Level: ______
Recommended Zone: ______
Scoring Rubrics
Data Risk (1-3)
| Score | Criteria |
|---|---|
| 1 (Low) | Public data only, no PII, no customer information |
| 2 (Medium) | Internal data, employee PII, non-regulated business data |
| 3 (High) | Customer PII, financial data, regulated information (NPI, PHI) |
Regulatory Risk (1-3)
| Score | Criteria |
|---|---|
| 1 (Low) | No regulatory applicability, internal productivity only |
| 2 (Medium) | FINRA 3110 supervision applies, internal recordkeeping |
| 3 (High) | SEC 17a-3/4, GLBA 501(b), customer-facing, examination scope |
Simplified Assessment Path
Agents may qualify for a simplified 1-page checklist assessment if ALL of the following apply:
- Total weighted score ≤ 1.5
- User count ≤ 10
- Not customer-facing
- No regulated data access
Otherwise, use the comprehensive assessment template.
Residual Risk Assessment
After applying mitigating controls, assess residual risk:
| Category | Initial Risk | Mitigating Controls | Residual Risk |
|---|---|---|---|
| Data | |||
| Regulatory | |||
| Operational | |||
| Security | |||
| Reputational |
Residual Risk Acceptable: Yes / No
If No, additional controls required:
Approval
Zone 2 Approval
| Role | Name | Signature | Date |
|---|---|---|---|
| Manager | |||
| AI Governance Lead |
Zone 3 Approval
| Role | Name | Signature | Date |
|---|---|---|---|
| Business Owner | |||
| AI Governance Lead | |||
| Compliance Officer | |||
| CISO | |||
| General Counsel (if customer-facing) |
Review Schedule
| Review Type | Frequency | Next Review Date |
|---|---|---|
| Risk reassessment | Annual | |
| Control effectiveness | Quarterly | |
| Bias testing | Quarterly |
Attachments
- Business case documentation
- Data flow diagram
- Test results
- Security assessment
- Bias testing results (Zone 3)
Last Updated: January 2026 FSI Agent Governance Framework v1.2