• Skills over pedigree: We hire for capability and outcomes, not for brand names or keyword matches.

• Human in the loop: AI assists decision-making, but humans always make the final call.

• Explainability: Every automated rejection must include a clear, plain-language reason code.

• Transparency: Both employers and candidates disclose when and how AI tools are used.

• Accountability: All AI tools used in hiring are regularly bias-audited to ensure fairness and accuracy.

We pledge to:

• Publish skills-based job descriptions with measurable proficiency levels.

• Guarantee a minimum 20% human review rate for all AI-screened applications.

• Provide rejection reason codes and a one-click appeal process reviewed by humans.

• Implement explainable AI models and retain audit logs for a minimum of two years.

• Replace long, unpaid take-home projects with short, compensated work trials.

• Conduct quarterly bias and fairness audits of all AI hiring systems.

• Permit declared AI assistance for résumés and portfolios while requiring verification and reproducibility of all submitted work.

• We believe every candidate deserves dignity, clarity, and an equitable chance to demonstrate their value.

  We therefore commit to:

  • Offer visibility into where and how AI is used in recruitment.

  • Honor transparency from applicants who disclose AI-assisted materials.

  • Evaluate reasoning, creativity, and communication — not automation skills alone.

  • Promote ethical AI use through training, mentorship, and apprenticeship pathways.

To protect the integrity of the U.S. white-collar workforce, we encourage:

• A National Skills Framework with standardized, verifiable credentials.

• Reskilling incentives for workers displaced by automation or AI.

• A federally protected Right to Human Review for all automated employment decisions.

• Annual public reporting of bias and false-negative rates for hiring AI systems.

Together, we can ensure that AI augments — not erodes — the future of American work.

By signing this pledge, our organization commits to ethical, transparent, and accountable AI-driven hiring practices. We recognize our shared responsibility to shape a labor market that values human potential and trust as much as technological progress.

Signed,

Organization / Representative
Date: ______________________

Here’s a practical, ethics-first playbook that helps employers, job-seekers, and the broader U.S. economy—and cools the ATS “arms race.”

1) Principles to anchor everything

• Skills over pedigree. Hire for demonstrable capabilities and outcomes, not brand names or keyword stuffing.

• Human in the loop. AI can screen; humans must decide.

• Explainable & contestable. Every automated rejection should have a plain-English reason and an appeal path.

• Transparent AI use—on both sides. Candidates and employers disclose how AI was used.

2) What employers & ATS vendors should implement (now → next 12 months)

A. Job design & postings

• Write skills-based roles: 6–8 core competencies, proficiency levels, and success criteria (e.g., “build a P&L forecast in Power BI with X, Y, Z data sets”).

• Show your process: stages, timelines, assessment types, and the % automated vs human review.

• Pay transparency & range to reduce noise applications and increase trust.

B. Fair screening configuration

• Human review floors: e.g., minimum 20% of applicants randomly sampled for human review regardless of AI score.

• Candidate-friendly cutoffs: avoid hard keyword gates; use multi-signal scoring (skills, portfolio, verified certs, work samples).

• Deduplicate with provenance: track repeated/AI-spam applications by identity verification—not by risky fingerprinting.

C. Explainability & appeals

• Rejection codes returned to candidates (e.g., “Insufficient SQL depth for the required assessment”).

• One-click appeal to upload evidence (portfolio link, brief Loom demo, or code sample) reviewed by a human within 5–7 days.

• Adverse-impact checks: quarterly bias audits; publish summary metrics.

D. Assessments that reduce cheating and noise

• Short, scoped work trials (60–90 minutes, paid if >60 min).

• Project debriefs over take-homes: candidates discuss HOW they solved prior work; interviewers probe reasoning.

• Simulated, monitored tasks (browser-based sandbox) that allow assisted but disclosed AI use (see “Candidate Declaration,” below).

E. Anti-gaming without witch hunts

• Content provenance: encourage portfolios with signed commits (Git), doc version history, or employer-verified accomplishments.

• AI-assist allowed, plagiarism not: run originality checks on work samples; compare reasoning vs output.

• Policy clarity in every posting: what AI assistance is allowed at each stage.

F. Vendor requirements (put these in your ATS contract)

• “Explainability API” to deliver rejection reasons + feature importance.

• “Human-override mode” and configurable human-review floors.

• “Bias dashboard” (selection rates, false negatives, subgroup analyses).

• “Audit log retention” (≥2 years) for regulatory or candidate challenges.

3) What job seekers should do (ethically & effectively)

A. Make your signal legible

• Skills-first résumé: bullets that show action → tools → outcome → metric.

• Portfolio proof: 2–4 case studies with data/code snippets, dashboards, or before/after business metrics.

• Verifiable breadcrumbs: GitHub commits, published dashboards (blurred/sample), letters confirming outcomes, or credential IDs.

B. Use AI—just disclose it

• Declare your AI assist: add a line “This résumé was drafted with AI and reviewed/edited by me. All claims are accurate and verifiable.”

• Practice “show your work”: be ready to re-create or live-explain any artifact you submit.

C. Apply smarter

• Fewer, better applications: tailor to the posted competencies; attach a 60-second Loom explaining fit.

• Network + referrals: pair every application with one targeted outreach to a hiring team member referencing the posted success criteria.

4) Lightweight policies & templates you can copy

A. Candidate AI-Use Declaration (attach to résumé)

I used AI tools to draft or proofread portions of my materials. All information reflects my real skills and experience. Any included work samples are my own or clearly labeled as collaborative. I can reproduce or explain any artifacts on request.

B. Job posting AI & assessment policy (employer)

We use AI to help organize applications; humans make hiring decisions. At least 20% of applications receive human review. If you’re rejected, we’ll provide a reason code and a one-click appeal. AI assistance is permitted for the résumé; the live exercise must reflect your own reasoning.

C. Rejection reason codes (examples)

• R-101: Core tool proficiency below posted level (SQL window functions).

• R-205: Missing required domain artifact (credit-risk dashboard).

• R-307: Assessment showed heavy copy-paste with insufficient explanation.

5) Policy actions for governments & standards bodies (to stabilize the market)

Near term (0–6 months)

• NIST HR-AI risk profile (fast-track): publish baseline guardrails for screening, explainability, and audit logs.

• Disclosure rule: employers must state where AI is used in hiring and provide rejection reason codes.

Mid term (6–18 months)

• Bias & error-rate reporting: large employers report annual adverse-impact and false-negative rates for automated screening.

• Incentives for apprenticeships/returnships in AI/analytics (tax credit per seat), with standardized outcomes tracking.

• Public skills taxonomies (build on O*NET) so postings and résumés share a common, machine-readable language.

Longer term (18–36 months)

• Right to human review for automated rejections (mirrors global best practice).

• Skills wallets / verifiable credentials: a national, privacy-preserving standard for certs and work history (reduces fraud and noise).

• Wage-insurance pilots for displaced white-collar workers transitioning via reskilling.

6) “Fair Hiring Stack” (one-page reference architecture)

1. Structured posting (skills + proficiency + success criteria)

2. ATS with explainability (rejection codes, human-review floor, audit logs)

3. Short, paid, monitored work trials allowing declared AI assist

4. Portfolio provenance (version history, credential verification)

5. Transparency loop (appeals, quarterly bias checks, public summaries)

7) Rollout plan for a real company (90-day sprint)

Weeks 1–2

• Convert 10 priority roles to skills-based templates; publish AI-use policy.

• Configure ATS: human-review floor, reason codes, appeal button.

Weeks 3–6

• Replace take-homes with 60–90 min paid trials; train interviewers on probing reasoning.

• Launch bias dashboard; set monthly review with HR + Legal.

Weeks 7–12

• Publish first transparency report (process, metrics, improvements).

• Partner with a local bootcamp/community college for 10-seat apprenticeship pipeline.

Why this works

• Reduces noise (structured signals + short trials)

• Restores trust (disclosure, reasons, appeals)

• Widens access (skills over pedigree, apprenticeships)

• Improves matching quality (human judgment where it matters)

• Future-proofs compliance (logs, bias audits, explainability)

In alignment with the values outlined in this pledge, we disclose that this document was collaboratively written and refined using artificial intelligence to demonstrate the transparent, ethical use of AI in content creation. All research, analysis, and language recommendations were generated and verified by the Voice for Change Foundation’s AI research tools under human supervision.

📄 Download the Fair Hiring & AI Transparency Employer Pledge (PDF)
🔗 View the Source Research and Policy Framework