AI in hiring: discrimination risks, EEOC rules, and HR best practices

Executive overview

AI tools can speed up hiring but carry serious legal risk if used carelessly. Biased training data, exclusionary filters, and automated rejections can violate the ADA, ADEA, and Title VII — and employers are liable even when the discrimination comes from third-party software.

The EEOC has issued guidance classifying algorithmic hiring tools as employment selection procedures subject to anti-discrimination law. The White House has published an eight-principle framework for ethical AI deployment.

HR is the last line of defence: no third-party tool shields an employer from liability.

Pros of AI in HR workflows

• Screens large applicant volumes quickly, reducing cost-per-hire
• Automates payroll, benefits, and compliance tasks, cutting human error
• Personalises onboarding via virtual reality and adaptive learning
• Optimises scheduling and generates meeting recaps automatically

How AI introduces discrimination

• Historical data bias: AI trained on past hiring data replicates existing patterns (e.g. Amazon's scrapped hiring tool favoured men)
• Exclusionary filters: keyword requirements can screen out qualified candidates from underrepresented groups
• Automated rejections: denying in-person interview alternatives violates ADA reasonable-accommodation requirements
• Asking about disabilities or medical conditions before a job offer is independently discriminatory

EEOC guidance and legal framework

• EEOC appointed its first Chief AI Officer in June 2024
• 2022 AI Bill of Rights established worker rights: protection from biased systems, data agency, and the right to opt out
• 2023 AI Disparate Impact Guidance: algorithmic hiring tools are subject to Title VII; employers must show tools are job-related and consistent with business necessity
• Four-Fifths Rule: if a selection rate for one group is less than 80% of the highest-performing group's rate, discrimination is indicated — though critics argue accepting up to 20% disadvantage is itself unethical

White House principles for AI in the workplace (May 2024)

1. Worker empowerment — workers have genuine input in AI design and oversight
2. Ethical development — systems protect workers by design
3. Governance and human oversight — clear internal procedures and evaluation processes
4. Transparency — disclose AI use to employees and job seekers
5. Labor rights protection — AI must not undermine workers' legal rights
6. Worker enablement — AI assists rather than replaces human judgment
7. Transition support — upskill workers displaced by AI
8. Responsible data use — worker data used only for legitimate business purposes

Best practices for HR

• Use knock-out questions (work authorisation, experience thresholds, relocation) instead of AI filters for early screening
• Bring in employment lawyers or AI experts before implementing new algorithmic tools
• Document policies and procedures for all AI-assisted employment processes
• Self-audit regularly: test outputs, retain records, and analyse impact on protected groups
• Train all users on ethical AI use
• Disclose AI use to candidates in job descriptions
• The No Third Party Shield principle: employers remain liable for what any platform does on their behalf