Free AI Risk Assessment Template Generator

AI Risk Assessment: Customer support reply drafting

Owner: Support Operations
Risk tier: Moderate (9/16)
Summary: Needs documented controls, owner sign-off, and scheduled review.

Scope
- Use case: Customer support reply drafting
- Data classification: Internal business data
- Exposure: Team workflow
- AI autonomy: Human reviewed output
- Audience: Customers or public

Required controls
- Document the use case, intended users, forbidden uses, and success criteria.
- Assign an accountable business owner and a technical owner for production changes.
- Require human review before launch and after any major prompt, model, data, or workflow change.
- Log model inputs, outputs, user feedback, and incidents with retention limits.
- Add customer-facing disclosure where AI materially shapes the output or decision.
- Create an escalation path for users to request human review or correction.

Risk register
1. Incorrect or unsupported output
   Impact: Users may act on flawed recommendations.
   Mitigation: Add source checks, confidence labels, and mandatory human review for edge cases.
2. Unclear accountability
   Impact: Issues may go unresolved after launch.
   Mitigation: Assign Support Operations as accountable owner with a monthly review task.
3. Data leakage
   Impact: Sensitive business or personal information could be exposed.
   Mitigation: Restrict inputs, mask sensitive fields, and limit logs to approved retention windows.
4. Customer misunderstanding
   Impact: Users may over-trust AI output or miss its limitations.
   Mitigation: Use clear disclosure, plain-language limitations, and a path to human support.

Approval checklist
- Business owner approved
- Privacy review complete
- Security review complete
- Human review process documented
- Monitoring dashboard or review process active
- Incident response owner named

Review cadence
- Review monthly until risk is stable, then quarterly.