Algorithmic Hiring Law Compliance 2026: US and Global Rules for AI Recruitment Tools
Five years ago, no US law specifically regulated AI hiring tools. Today there are active laws in New York City, Illinois, Maryland, Colorado — plus EU enforcement starting August 2026. If your product touches employment decisions, the compliance burden has arrived.
The Algorithmic Hiring Law Landscape in 2026
The regulatory environment for AI hiring tools has changed dramatically since 2020. What was once a legal gray area — AI can score candidates, no specific law says it can't — is now a multi-layer compliance requirement spanning local, state, and international law.
The common thread across all these laws: transparency and accountability. Regulators are not (yet) banning AI hiring tools. They are requiring that companies disclose their use, audit for bias, obtain consent, and give candidates procedural rights. The framework mirrors the arc of data privacy law — permissive initially, then disclosure-and-consent requirements, then substantive restrictions as harm evidence accumulates.
US Laws: Active and Enforceable
NYC Local Law 144
New York City
Scope: Automated Employment Decision Tools (AEDTs) used to screen candidates for NYC-based positions.
Key requirements:
- ▸Independent third-party bias audit within 1 year of each use
- ▸Public posting of bias audit results (selection rates and impact ratios by race/ethnicity and sex)
- ▸10 business days' advance written notice to candidates that an AEDT will be used
- ▸Offer an alternative selection process for candidates who opt out
- ▸2-year record retention for audit results and candidate notices
Penalties: $375–$1,500 per violation per day. NYC DCWP enforces.
Who it applies to: HR tech vendors and NYC employers using AI to screen candidates for NYC jobs.
Illinois AI Video Interview Act
Illinois (statewide)
Scope: AI tools that analyze video interview recordings to evaluate applicants for employment in Illinois.
Key requirements:
- ▸Notify applicants before the interview that AI will analyze the video
- ▸Explain how the AI works and what characteristics it evaluates
- ▸Obtain applicant consent before AI analysis
- ▸Do not share AI analysis of applicant videos with any person except qualified experts needed to evaluate the AI
- ▸Destroy video recordings on applicant request within 30 days (or within 1 year of last use if applicant doesn't request deletion)
Penalties: Private right of action. Statutory damages for violations. IDHR (Illinois Dept of Human Rights) enforcement.
Who it applies to: Any company using AI video interview analysis tools with Illinois applicants — including remote roles filled from Illinois.
Maryland HB 1202
Maryland (statewide)
Scope: Employers using facial recognition technology during job interviews with Maryland applicants.
Key requirements:
- ▸Obtain applicant's written consent before using facial recognition during a job interview
- ▸Cannot require consent as a condition of employment application — applicant can decline
Penalties: Civil penalties enforced by Maryland Attorney General.
Who it applies to: Narrower scope: applies to facial recognition specifically, not all AI interview tools. Relevant for video interview platforms using face analysis.
Colorado SB 205
Colorado (statewide)
Scope: High-risk AI systems, including AI used in employment decisions: hiring, firing, performance evaluation, compensation, and promotion.
Key requirements:
- ▸AI developers must use reasonable care to protect Coloradans from algorithmic discrimination
- ▸Deployers (employers) must disclose to employees and applicants that high-risk AI is being used
- ▸Provide a meaningful human review process for adverse AI decisions
- ▸Post on public website what high-risk AI systems are deployed and their general purpose
- ▸Annual impact assessments to detect algorithmic discrimination
- ▸Accommodate disability-based requests for alternative processes
Penalties: Colorado AG enforces. No private right of action (unlike BIPA). Penalties under Colorado Consumer Protection Act.
Who it applies to: Both AI tool developers (vendors) and employers (deployers) using AI in employment decisions affecting Colorado residents.
EU AI Act: Employment AI Is High Risk
Under the EU AI Act, AI systems used in employment, worker management, and access to employment are classified as high-risk AI (Annex III). This means full compliance requirements apply — not just transparency disclosures. High-risk AI obligations become enforceable August 2, 2026.
Covered employment AI under the EU AI Act includes:
- AI for recruitment and candidate screening — filtering applications, ranking candidates, assessing interview performance
- AI for promotion decisions — tools that evaluate employees for advancement
- AI for task allocation and monitoring — AI workforce management platforms
- AI for performance evaluation and termination decisions
For EU high-risk employment AI, compliance requires: a full conformity assessment, technical documentation (a "technical file"), a risk management system, human oversight mechanisms, audit logs, data governance documentation, and registration in the EU AI database. This is a significant compliance burden — closer to a medical device regulatory process than a software deployment.
Federal Law: EEOC and Title VII Still Apply
Even where no specific algorithmic hiring law applies, federal employment discrimination law does. The EEOC has issued guidance making clear:
Disparate Impact Applies to AI
Title VII's disparate impact theory applies to AI hiring tools. If an AI screening system produces selection rates that are significantly lower for protected classes (race, sex, national origin) without business necessity justification, it can be unlawful employment discrimination.
Employer Liability Is Non-Delegable
An employer cannot avoid Title VII liability by outsourcing hiring decisions to an AI vendor. If your vendor's AI produces illegal disparate impact, you — the employer — are responsible. 'The AI did it' is not a defense.
ADA and Disability Accommodations
AI hiring tools may screen out individuals with disabilities in violation of the ADA. The EEOC's 2022 AI guidance specifically notes that predictive AI tools that fail to account for disability-related limitations may trigger ADA liability. Employers must be able to explain what their AI measures and provide accommodations.
OFCCP for Federal Contractors
Federal contractors are subject to OFCCP oversight, which has expanded scrutiny of algorithmic hiring tools. Federal contractors using AI in hiring are at elevated regulatory risk.
Compliance Matrix: What You Must Do by Jurisdiction
Use this matrix to understand what specific obligations apply based on where you operate and what your product does:
If: You sell an AI resume screening or candidate ranking tool
- ▸NYC LL144: Get a bias audit if your customers hire for NYC positions
- ▸Colorado SB 205: Use reasonable care to prevent algorithmic discrimination; provide disclosures
- ▸EU AI Act: Full high-risk AI compliance for EU customers (Aug 2026)
- ▸EEOC/Title VII: Conduct disparate impact analysis; document business necessity
If: You sell an AI video interview platform
- ▸Illinois AI Video Interview Act: Require customers to notify applicants and get consent; build deletion flow
- ▸Maryland HB 1202: If you use facial recognition, require customer consent workflows
- ▸NYC LL144: If facial scoring influences hiring decisions, may be an AEDT requiring audit
- ▸BIPA: If you extract face geometry or voice biometrics, BIPA consent and policy required for Illinois users
If: You are an employer using AI tools to screen candidates
- ▸NYC LL144: Confirm vendor bias audit exists and is current; notify NYC candidates; offer alternative process
- ▸Colorado SB 205: Disclose AI use; provide human review for adverse decisions; accommodate disability requests
- ▸Illinois Video Interview Act: Ensure vendor consent and notification flows are in place
- ▸EEOC: Monitor for disparate impact; be able to justify AI decisions if challenged
If: You build AI for promotion, performance review, or compensation decisions
- ▸Colorado SB 205: High-risk AI classification; disclosure, human review, annual impact assessment
- ▸EU AI Act: High-risk employment AI — full conformity assessment required for EU use
- ▸EEOC: Promotion and compensation decisions subject to Title VII disparate impact analysis
- ▸OFCCP: Federal contractors face additional scrutiny
Pending Legislation to Watch
California AB 331 — Automated Decisions Systems
Stalled but activeWould require impact assessments and disclosure notices for high-risk automated decision systems including employment AI in California. Has passed committee multiple times but not yet signed into law. Watch for reintroduction.
Federal AEIOU Act (Algorithmic Employment in Our Organizations)
Introduced 2025Would require employers with 15+ employees to disclose AI use in hiring to applicants, conduct and publish bias assessments, and provide human review for automated rejections. Has support from EEOC commissioners but not yet passed.
New York State Legislation
Multiple bills activeAlbany has seen multiple bills to expand NYC Local Law 144-style requirements statewide. Watch for a statewide NY AEDT law that would extend NYC-style obligations to all NY employers, not just NYC-based hiring.
Washington SB 5116 — Artificial Intelligence
Passed 2024, amendments pendingWashington's broad AI accountability law covers automated employment decisions. Requires developers to perform impact assessments and deployers to disclose AI use to affected persons. Implementation details still being refined through rulemaking.
Algorithmic Hiring Compliance Checklist for HR Tech Vendors
Audit your HR tech product's compliance posture
RatedWithAI helps tech teams track compliance requirements across accessibility, privacy, and AI regulation. Start with a free scan to understand where you stand.
Scan Your Product for Free →Frequently Asked Questions
Does my AI hiring tool need a bias audit even if I'm not in New York?
Yes, if your customers use your tool to screen candidates for NYC-based positions. NYC Local Law 144 applies based on where the job is located — not where the employer or vendor is headquartered. If you sell to national employers who hire in NYC, your tool may be subject to LL144 audit requirements even if you've never operated in New York.
What's the difference between algorithmic hiring laws and anti-discrimination law?
Traditional anti-discrimination law (Title VII, ADEA, ADA) prohibits discriminatory outcomes but was enacted before AI existed. Algorithmic hiring laws are a new layer that specifically regulates AI processes — requiring audits, disclosures, and consent regardless of whether discrimination actually occurs. You can comply with algorithmic hiring laws while still facing Title VII liability (if your compliant tool still produces disparate impact), and vice versa.
Can I rely on my AI vendor to handle compliance, or do I (as the employer) need to act?
Both. Vendors must conduct bias audits and post results publicly. Employers must verify vendor compliance, notify candidates, offer alternative processes, and retain records. Employment discrimination liability under federal law cannot be delegated to a vendor — the employer using the tool is always responsible for ensuring it doesn't produce discriminatory outcomes. Always ask vendors for their LL144 audit link before deploying in NYC.
Does the Illinois AI Video Interview Act apply to remote roles that might be filled by Illinois residents?
Courts have interpreted Illinois employment laws broadly. If a role is open to applicants who may be Illinois residents, and you're using AI video analysis during the interview process, applying the Illinois Video Interview Act is the conservative and legally safer position. Practically: build consent and disclosure into your platform for all video interviews, not just explicitly Illinois-targeted ones.
What is a 'bias audit' and how long does it take to get one?
A bias audit under NYC Local Law 144 is an analysis conducted by an independent third party that calculates selection rates and impact ratios for your AI tool across sex, race/ethnicity, and intersectional categories using real or representative historical data. Bias audits from specialized firms typically take 4–12 weeks depending on data availability and tool complexity. Firms specializing in AEDT audits include Bayes Impact, O'Neil Risk Consulting, and Booth/Zhu Research — though the market is growing rapidly.