Sheet 07 · ART-06
Responsible-use review checklist
The questions a reviewer works through, with the answers recorded. It exists so reviews are consistent between reviewers and legible to the requester afterward.
Artifact details
- Format
- Worksheet
- Owner
- AI Program Manager coordinates; each function answers its own section.
- When it is used
- During structured and elevated review; attached to the decision record.
- Retention
- Attached to the decision record.
Contributors: Information Security, Risk, Legal & Privacy, Data & Reporting, People & Culture
Downloads
The blank template is a standalone worksheet with every field and its guidance, ready to fill in or print. Printing uses this page’s print stylesheet, which drops the navigation and the download controls.
Required fields
The guidance matters more than the field name. A field labelled “risks” with no guidance gets filled in with “standard AI risks” and the artifact stops being worth reading.
| Field | Guidance | Required |
|---|---|---|
| Problem and alternatives | Is AI the right response, or would a process fix work? | Required |
| Data | What goes in, where it is processed, who else can see it, how long it is kept. | Required |
| Output and consequence | Who is affected if it is wrong, and how would anyone find out? | Required |
| Human oversight | Named role, specific check, defined frequency, authority to override. | Required |
| Fairness and affected groups | Only where people are affected. What evidence exists, and who validated it? | Optional |
| Transparency | Do the people affected know, and does the firm need to say so? | Required |
| Failure and reversal | What happens when it is wrong, and what is the undo path? | Required |
| Ongoing monitoring | What is watched, by whom, how often, and what triggers a stop. | Required |
Worked example
Filled in against the sample record set, so the template can be judged on what a real entry looks like rather than on its headings.
UC-004 — Recruiting screening, review record
- Problem and alternatives: capacity constraint is real. Two distinct activities are consuming it — evaluation and coordination. Coordination is the larger share and carries far less risk. Recorded as the key finding.
- Data: applicant resumes, contact details, interview notes. Restricted. Vendor default retention exceeds the firm's recruiting schedule — gap identified.
- Output and consequence: a ranked shortlist would shape who gets interviewed. A candidate who is not advanced would have no explanation available.
Findings that decided it
- Human oversight: reviewing a ranked list inherits the ranking. Reviewers rarely reorder a sorted list, so the proposed oversight would not have functioned as a control.
- Fairness: vendor could not produce adverse-impact testing. Firm has no internal capability to validate independently. This was the finding that settled the question.
- Transparency: applicant privacy notice does not cover automated processing.
- Failure and reversal: no reversal path for a candidate screened out — they simply never hear back.
Outcome
- Screening scope declined (DEC-2026-04). Reduced coordination scope approved with conditions (DEC-2026-05).
- Reviewers: People & Culture, Risk, Information Security, two hiring managers, AI Governance Committee.
Blank template
The same fields with nothing in them. Print this page, or use the download above to get a standalone file.
Problem and alternatives
Is AI the right response, or would a process fix work?
Data
What goes in, where it is processed, who else can see it, how long it is kept.
Output and consequence
Who is affected if it is wrong, and how would anyone find out?
Human oversight
Named role, specific check, defined frequency, authority to override.
Fairness and affected groupsoptional
Only where people are affected. What evidence exists, and who validated it?
Transparency
Do the people affected know, and does the firm need to say so?
Failure and reversal
What happens when it is wrong, and what is the undo path?
Ongoing monitoring
What is watched, by whom, how often, and what triggers a stop.