You get:
- dense legal paragraphs no employee reads
- vague language that can’t be enforced (“reasonable,” “appropriate”)
- no clear consequences for violations
- policies that contradict each other
- no acknowledgment line (so you can’t prove they saw it)
But a policy is not a legal document for lawyers.
It is a rulebook employees must understand and follow.
- Purpose tells them why — compliance feels less arbitrary
- Scope tells them who — no confusion about applicability
- Actionable rules use “must” and “must not”
- Enforcement signals seriousness
- An acknowledgment line creates accountability
Without clarity, policies create liability instead of reducing it.
This framework forces AI to think like an HR compliance officer who writes for humans.
Assume the role of a policy drafting specialist and compliance writer who writes clear, enforceable rules for real workplaces. Your task is to draft a 1-2 page policy using plain English (8th-grade reading level). STRUCTURE: 1. PURPOSE (1 sentence) Why this policy exists — the problem it solves 2. SCOPE (1 sentence) Who this policy applies to (all employees, specific roles, contractors, etc.) 3. POLICY STATEMENTS (3-7 clear, actionable rules) Each rule must start with: - "Employees must..." (required action) - "Employees must not..." (prohibited action) - No vague words: "reasonable," "appropriate," "as needed" (unless defined) 4. EXCEPTIONS (if any) Specific situations where the rule does not apply 5. ENFORCEMENT (1-2 sentences) What happens if an employee violates this policy 6. EFFECTIVE DATE & REVIEW SCHEDULE 7. ACKNOWLEDGMENT LINE "I have read and understood this policy. Signature: ________ Date: ________" INPUTS: Policy Topic: [REMOTE WORK / EXPENSE REIMBURSEMENT / DATA SECURITY / AI USAGE / CODE OF CONDUCT / OTHER] Company Size: [1-10 / 11-50 / 51-200 / 200+] Industry: [INSERT INDUSTRY] Specific Concerns (optional): [WHAT KEEPS YOU UP AT NIGHT ABOUT THIS TOPIC?] RULES: - No legalese — write for an 8th-grade reading level - "Must" and "must not" are stronger than "should" or "encouraged" - Define any word that could be interpreted differently by different employees - Enforcement must be specific (e.g., "written warning," "termination," not "disciplinary action") - Add disclaimer: "This policy does not create a contract. [Company Name] reserves the right to modify or rescind."
- Have an HR professional or employment lawyer review any policy before implementation.
- The acknowledgment line is critical — keep signed copies in employee files.
- If a rule can’t be enforced, remove it (unenforceable policies create liability).
- Review all policies annually; outdated policies are as dangerous as bad ones.
- Write policies assuming an employee will try to find a loophole — close it.
Policy Topic: AI Usage (employees using ChatGPT, Copilot, Midjourney, etc. for work)
Company Size: 11-50
Industry: Marketing agency
Specific Concerns: “I’m worried employees will paste client confidential information into public AI tools without realizing it’s being used for training.”
This framework improves outcomes by forcing:
- 8th-grade reading level accessibility
- “must/must not” enforceability
- defined terms (no vagueness)
- specific enforcement consequences
- acknowledgment line for accountability
Great policies don’t just tell employees what not to do — they make compliance obvious.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI legal tools, policy drafting frameworks, and practical strategies for professionals and business owners.