CopilotForgeDraft

Templates › AI risk assessment

AI risk assessment

Template id: ai-risk-assessment

Responsible skills: security-privacy:data-exposure-review, security-privacy:connector-review and security-privacy:privacy-assessment. Section 1 records the risk tier set during opportunity risk classification.

Every fact the organisation must supply is marked [To confirm: ...]. Example rows are fictional. Each finding is classified blocking or advisory, names the risk, and gives the fix and owner. From risk tier 2 a blocking finding stops release and is not traded against dates. Disagreement escalates to the accountable executive, whose risk acceptance is recorded in section 7. Legal questions end with the Data Protection Officer.

1. Solution and risk tier

FieldEntry
Solution name[To confirm]
Brief or design reference[To confirm]
Business owner[To confirm]
Audience and sharing scope[To confirm]
Risk tier (0-4) and rationale[To confirm]
Reviewers and date[To confirm]

2. Data exposure

Completed with security-privacy:data-exposure-review. Grounding is permission-trimmed, so overshared content becomes discoverable; judge whether the permissions are right, not only whether they are respected.

Source or output pathWho can access todayIntended audienceSensitivity or labelWeb or public?Gap
[To confirm][To confirm][To confirm][To confirm]Yes / No[To confirm]

Permission-trimming test evidence: [To confirm: reference and user profiles used]

3. Connectors and actions

Completed with security-privacy:connector-review.

Connector, action, tool, MCP server or APIOperations usedIdentity (user, maker, service)Permissions grantedData policy and zoneData leaves tenant?Confirmation for writes
[To confirm][To confirm][To confirm][To confirm][To confirm]Yes / NoYes / No / N/A

4. Privacy

Completed with security-privacy:privacy-assessment.

QuestionEntry
Whose personal data, which categories (special category?)[To confirm]
Personal data map: grounding, input, output, storage, onward flows[To confirm]
Purpose and compatibility with original collection[To confirm]
Lawful basis, as confirmed by the DPO, with date[To confirm]
Transparency (notices and in-product statement)[To confirm]
Minimisation measures[To confirm]
Retention for prompts, responses, transcripts and logs, with owner[To confirm]
Rights handling covers AI data[To confirm]
DPIA screening result and DPIA reference if required[To confirm]
Outputs inform decisions about individuals? Human review design[To confirm]

5. Findings

IDAreaFinding and riskRule or standardBlocking or advisoryFixOwnerDueStatus and evidence
Fictional example F1Data exposureSalary files shared with all staff are reachable through the agentContent shared beyond intended audienceBlockingRemove broad grant at source; re-run permission testsFictional example: HR site owner[To confirm]Open
[To confirm][To confirm][To confirm][To confirm]Blocking / Advisory[To confirm][To confirm][To confirm][To confirm]

6. Verdict

AreaVerdict
Data exposurePass / Pass with advisory findings / Blocked
Connectors and actionsPass / Pass with advisory findings / Blocked / N/A
PrivacyPass / Pass with advisory findings / Blocked / N/A
OverallPass / Pass with advisory findings / Blocked

7. Risk acceptance (only where a finding is disputed)

Finding IDRisk acceptedRationaleAccountable executiveDateReview date
[To confirm][To confirm][To confirm][To confirm][To confirm][To confirm]

Credit: Template from CopilotForge by Simon Foote (https://github.com/SimonFoote-greensparx/CopilotForge), licensed under CC BY 4.0. Keep this credit when you share or adapt the template; in your own completed document you may keep it as a footer or remove it.

Download the template (Markdown)