Table of Contents
Quick Answer
HR and recruiting AI prompts in 2026 must balance speed (faster hiring), fairness (bias-free), and compliance (EEOC, GDPR, India DPDP). The best prompts explicitly exclude protected attributes and anchor to measurable skills.
- Never infer gender, age, ethnicity, or nationality from names or locations
- Anchor to skills + evidence, not "culture fit"
- Log every AI-assisted decision for audit trail (EU AI Act requires it)
Prompt Examples
Write a job description for a [role] at [company]. Required skills: [list]. Nice-to-have: [list]. Mission: [paste]. Format: role overview (60 words), what you'll do (5 bullets), what we're looking for (5 bullets), compensation range, benefits. Remove gendered language (check with a tool). Include inclusive boilerplate.
Screen these 20 resumes for [role]. Criteria: [list 5 must-haves]. For each candidate: score 1-5 on each must-have, one-sentence evidence, overall recommendation (interview / pass / maybe). Never reference name, gender, age, photo, or location.
Generate 10 interview questions for [role]. Mix: 3 behavioral (STAR format), 3 technical, 3 situational, 1 values-based. For each: what it tests, what a great answer looks like, red flags.
Draft a LinkedIn sourcing message for a Senior [role] candidate. Company: [brief]. Why this role: [brief]. Don't praise them generically — reference 1 specific thing from their profile. Under 300 chars. No "I came across your profile".
Review this candidate's take-home assignment: [paste]. Rubric: [paste]. For each rubric criterion: score + 1-sentence justification citing the submission. End with hire / don't hire / maybe + confidence 1-5.
A candidate asked about [topic, e.g., remote policy, salary negotiation]. Our company's stance: [paste]. Draft a reply that is honest, brand-on, and under 100 words. Don't promise anything not in our policy.
Draft rejection emails for 3 scenarios: (1) rejected after resume screen, (2) rejected after phone screen, (3) rejected after final. Each under 80 words. Honest, kind, specific to the stage. No "we'll keep your resume on file" unless we will.
Generate a 30/60/90 day plan for a new [role]. Include: 30-day goals (learn), 60-day (contribute), 90-day (own). Each with 3 concrete milestones and success metrics. Manager check-in cadence.
Write an offer letter for [role] at [company]. Comp: [details]. Start: [date]. Include: at-will language (US) or relevant jurisdiction language, equity/RSU terms, benefits summary, conditions (background check, references). Plain English summary at top, legal terms below.
Draft a performance review for an employee. Context: [paste]. Goals from last cycle: [paste]. Achievements: [paste]. Areas of growth: [paste]. Format: summary, strengths (3), growth areas (2), goals for next cycle (3). Tone: honest, development-focused, not punitive.
How to Customize
- Feed job description + rubric at the start of screening sessions — consistency
- Explicitly tell the model to ignore protected attributes
- For comp questions, paste your comp bands — AI shouldn't guess
- Always keep human review on final hire/fire decisions (EU AI Act high-risk)
Common Mistakes
- Letting AI make hiring decisions alone — high-risk under EU AI Act
- Not disclosing AI screening to candidates — required in NYC, EU, coming to more regions
- Culture fit prompts — proxy for bias
- Using consumer ChatGPT for employee data — use Enterprise + DPA
Top Tools
Tool
Strength
Free Tier
Best Use Case
Paradox Olivia
Conversational sourcing
No
High-volume
HireVue
Structured interviews
No
Enterprise
Eightfold
Talent intelligence
No
Internal mobility
Gem
CRM + AI outreach
Yes
Sourcing teams
ChatGPT Team
Flexible prompts
No
Small HR teams
FAQs
Is it legal for AI to screen resumes? In most US states yes with disclosure; NYC AEDT Law requires bias audit; EU AI Act classifies it as high-risk in 2026. Disclose + audit.
Can AI conduct interviews? First-round screening yes (with disclosure). Final decisions must be human.
How do I avoid bias in AI hiring? Remove PII before screening, use structured rubrics, audit outputs quarterly for disparate impact.
Best model for HR tasks? Claude 4.6 for nuanced writing, GPT-5 for speed, Gemini for Workspace integration.
Can AI write performance reviews? Yes as a draft — final review must be from the manager with concrete examples they observed.
What about candidate data privacy? Under GDPR and India DPDP, get explicit consent for AI processing. Delete data within the posted retention window.
Does using AI hurt candidate experience? Not if transparent. Candidates appreciate faster responses; they resent being ghosted.
Conclusion
HR AI in 2026 must be fast, fair, and auditable. These 20 prompts get you there — JDs, sourcing, screening, interviews, offers, reviews.
Publishing your employer brand content? Launch a careers blog on Misar.Blog↗ — SEO-optimized for "jobs at [company]" queries.