AI HR and Recruiting Safety Checklist
AI Privacy Rule
Keep sensitive information out of general AI prompts, including names, family details, email addresses, phone numbers, account data, customer records, employee files, financial records, legal documents, medical information, and confidential business details. Use placeholders, redacted examples, or approved systems when needed, and keep human review before important actions. AI Privacy Rules
AI at HR / Recruiting / Step 4
Use This Checklist Before Using AI in HR or Recruiting
AI can help HR and recruiting teams draft, summarize, organize, and prepare. Before using AI-assisted output in any live workflow, confirm that the seven core requirements below are met. If any item cannot be confirmed, do not proceed until it is resolved.
The Pre-Use Checklist
- Workflow type: Is AI supporting job descriptions, candidate communication, interview prep, resume review support, onboarding, employee requests, or policy explanations?
- Decision risk: Could this output affect hiring, firing, promotion, compensation, discipline, accommodation, or employment outcomes? If yes, a qualified human must review it before use.
- Privacy: Does the input include candidate, employee, medical, financial, workplace, or sensitive personal information? If yes, confirm the tool is approved for this data type.
- Tool approval: Has the AI tool been approved by IT, legal, or HR leadership for this specific workflow and data category?
- Fairness: Has the output been checked for biased, exclusionary, inconsistent, or irrelevant language?
- Accuracy: Has the output been verified against approved HR, legal, benefits, policy, and role materials?
- Human review: Has the designated HR, recruiting, legal, or manager reviewer approved the final output?
How to Use This Checklist
Run this checklist at the beginning of any new AI-assisted HR workflow, not just once at onboarding. When you add a new use case — a new type of job description, a new employee communication category, or a new HR request template — treat it as a new checklist instance. Print it as a one-page reference, add it to your HR AI policy documentation, or build it into your team’s workflow approval process. The goal is consistent application, not occasional compliance.
When to Stop and Escalate
- The workflow could independently affect an employment decision without human review
- The AI tool has not been approved for the type of data the workflow involves
- The output contains language that cannot be verified against an approved policy source
- The output involves a sensitive situation: harassment, discrimination, leave, accommodation, discipline, or legal exposure
- The reviewer is not confident the output is accurate, fair, and compliant before it reaches employees or candidates
The Rule
Use AI to draft and organize, but keep people responsible for employment decisions, sensitive cases, policy interpretation, fairness, privacy, and final communication. This checklist is a gate — run it every time, not just the first time.
Where Safety Checklist Use Breaks Down
The most common failure mode is treating the checklist as a one-time onboarding step rather than a per-workflow requirement. Teams that run it once and assume all future AI use is covered end up with inconsistent review, undocumented approvals, and growing exposure as the range of AI use cases expands. Assign a named reviewer for each workflow category and make the checklist a required step in your AI approval process rather than an optional reference.
Quick Reference
- Run this checklist for every new HR or recruiting AI use case — not just once at setup
- All seven checklist items must be confirmed before AI output is used in a live workflow
- If any item cannot be confirmed, pause the workflow and escalate
- Assign a named human reviewer to every AI-assisted workflow that affects employees or candidates
- Document approvals, reviewers, and checklist results for compliance and audit purposes
Example in Practice: Running the Checklist on a New Use Case
The prompt: “We want to start using AI for this HR workflow: [describe it, e.g. ‘drafting interview confirmation emails from our scheduling tool’]. Walk it through these seven checks one by one: workflow type, decision risk, privacy, tool approval, fairness, accuracy, human review. For each check, state what we’d need to confirm and who on a typical HR team should own that confirmation. Mark any check that can’t be confirmed yet as a blocker.”
What you get back: A filled-in checklist draft with owners and open blockers — the difference between “we ran the checklist” and a documented record of who confirmed what before the workflow went live.
Check before using: The AI’s checklist walk-through is preparation, not approval — the named reviewer still confirms each item personally before the workflow is used.
Sources & Further Reading
- OWASP Top 10 for LLM Applications — the standard reference on data-disclosure and misuse risks the privacy and tool-approval checks are designed to catch.
- NIST AI Risk Management Framework — the govern/map/measure/manage structure this checklist condenses into a per-workflow gate.
Prompt Pack Resource
HR / Recruiting Prompt Pack Includes Safety-Ready Templates
The HR / Recruiting Prompt Pack includes reviewed prompts for job descriptions, candidate communication, interview guides, and resume review — each built with fairness and privacy in mind, ready to use as a starting point in your approved workflows.
Get the HR / Recruiting Prompt Pack