Dieser Artikel ist derzeit auf Englisch verfügbar.
AI in Hiring and Employment: A Compliance Guide for US Employers
US employers using AI in hiring, monitoring, or employment decisions face EEOC enforcement, NYC Local Law 144, state AI hiring laws, and growing plaintiff's bar attention. Here is the compliance framework.
Key Takeaways
EEOC's 2023 guidance establishes that employers are liable for discriminatory AI hiring outcomes regardless of whether the vendor designed the algorithm — you cannot outsource discrimination liability to your ATS or video interview vendor.
NYC Local Law 144 (in force since July 2023) requires employers with NYC employees using AEDTs to conduct annual independent bias audits, publish results publicly, and notify candidates 10 business days before AEDT use. Non-compliance penalties start at $375–$1,500 per violation.
Colorado SB 24-205 (effective 2026) applies to high-risk AI in employment decisions — requiring risk assessments, employee notification, and appeal rights. Employers with Colorado employees should begin compliance now.
Adverse impact analysis — measuring whether AI tools produce meaningfully different selection rates across demographic groups — is the primary tool for assessing discrimination risk. The EEOC's 4/5ths rule flags potential adverse impact when protected group selection rates fall below 80% of the highest-selected group.
Pre-deployment bias auditing by an independent third party is increasingly expected for any AI hiring tool at significant scale. Vendors who cannot provide bias audit results should be treated as higher risk.
Worker surveillance laws in Connecticut, New York, and Delaware require advance written notice of electronic monitoring. ADA requirements mean monitoring data cannot be used to infer disability status.
"Nur zu Informationszwecken. Dieser Artikel stellt keine rechtliche, regulatorische, finanzielle oder professionelle Beratung dar. Konsultieren Sie einen qualifizierten Spezialisten für spezifische Beratung."
The federal legal framework: what applies to US employers now
No single federal AI employment law exists in the United States, but three established federal laws create material compliance obligations for employers using AI in hiring, management, and termination decisions.
Title VII of the Civil Rights Act 1964 prohibits employment discrimination based on race, colour, religion, sex, or national origin. If an AI hiring tool produces significantly lower selection rates for a protected group — the four-fifths rule is a common benchmark, though not a safe harbour — that is evidence of unlawful disparate impact. The EEOC's May 2023 technical guidance on AI in employment confirmed employers are responsible for discriminatory AI outcomes even when using third-party vendor tools, and that disparate impact analysis applies fully to algorithmic systems.
The Americans with Disabilities Act creates two obligations. First, AI screening tools that screen out qualified candidates with disabilities — for example, video AI that disadvantages people with speech impairments — may create ADA liability. Second, employers must provide reasonable accommodations in the assessment process. If your AI interview tool is inaccessible to a candidate with a disability, you must offer an alternative.
The Age Discrimination in Employment Act (ADEA) protects workers 40 and older. AI tools can develop systematic age bias — filtering out longer career histories, penalising graduation years, or downgrading experience-heavy CVs. The ADEA authorises disparate impact claims under Smith v City of Jackson (2005), but applies a narrower standard than post-1991 Title VII — the Wards Cove framework, requiring plaintiffs to identify a specific employment practice causing adverse impact, with employers able to raise the “reasonable factors other than age” (RFOA) defence. Separately, Gross v FBL Financial Services (2009) governs ADEA disparate treatment claims, establishing a “but-for” causation standard. The EEOC continues to apply both to AI hiring tools.
State and city AI employment laws: where the action is
State and city laws are moving faster than federal law and create a patchwork compliance challenge for multi-state employers.
New York City Local Law 144 is the most operationally demanding. If you use automated employment decision tools (AEDTs) in hiring or promotion decisions for NYC-based roles, you must: commission an independent bias audit of the AEDT within the past 12 months; post audit results publicly on your website or a job listing website; and give candidates at least 10 business days' advance notice that an AEDT will be used, what it assesses, and that they can request an alternative process. Civil penalties are $500–$1,500 per day per violation. The December 2025 NYC Comptroller audit found DCWP enforcement had been limited, but noted DCWP committed to improving its enforcement program. Do not treat limited past enforcement as a prediction of future enforcement.
Illinois AI Video Interview Act requires notification and consent before using AI to analyse video interviews, a plain-language explanation of how the AI works, and deletion of videos and AI-generated data within 30 days of a candidate's request. Illinois employers face civil penalties for violations.
California does not yet have a comprehensive AI employment law, but CCPA/CPRA rights apply to employee personal data (post-2023 amendment), the Fair Employment and Housing Act (FEHA) applies the same disparate impact framework as federal law, and the CPPA is developing automated decision-making technology regulations that will affect employers.
Colorado SB 24-205 (the Colorado AI Act), originally slated to take effect June 30, 2026, is now the subject of a constitutional challenge by xAI and DOJ intervention. A replacement bill (SB 189) passed the Colorado legislature in May 2026 and was headed to the governor. Multi-state employers should monitor this closely.
Maryland, New Jersey, Vermont, and several other states have passed or are considering AI employment bills. The landscape is fragmenting. Centralised tracking of state AI employment laws is now a compliance necessity for any multi-state employer.
NLRA obligations: AI and worker organizing
The National Labor Relations Act protects employees' rights to organise, discuss wages, and engage in protected concerted activity regardless of whether a union is present. The NLRB has made clear that AI-enabled monitoring that identifies workers engaged in protected activity, or that is used to retaliate against organisers, violates the NLRA. Several 2024-2025 NLRB cases address algorithmic management in gig and delivery platforms. Employers using AI monitoring tools should audit whether those tools could be used in ways that chill protected activity.
Your EEOC compliance obligations
Before deploying an AI hiring or assessment tool, conduct a disparate impact analysis: does the tool select, screen, or score applicants of different protected groups at significantly different rates? Apply the four-fifths rule as a starting point — if a protected group's selection rate is less than 80% of the highest-selected group's rate, that is a potential indicator. This is not an automatic legal violation but it triggers the need for a business necessity justification.
Under the Uniform Guidelines on Employee Selection Procedures, validation studies demonstrating the predictive validity of the assessment tool are the primary defence to a disparate impact claim. Require vendors to provide validation studies. If a vendor cannot provide them, that is a significant risk signal.
The EEOC can investigate individual charges and systemic discrimination. In the AI context, systemic investigations — looking at patterns across many applicants — are more likely than individual cases. Maintain records of AI hiring decisions, selection rates by demographic group, and any bias audits you have conducted.
Practical compliance steps for US employers
Audit your current AI tools: list every AI tool used in hiring, performance management, and workforce decisions; identify the vendor; confirm what data it uses and what it outputs. For each tool, review the vendor's disparate impact testing — require this in your procurement process and in your vendor contracts. Implement a human review process for all consequential AI-assisted decisions. Train hiring managers on what AI tools assess and their limitations. Document your compliance analysis — if a charge is filed, documentation of good-faith compliance efforts is material to your defence. Review your job postings for NYC LL144 compliance if you post any roles open to NYC-based candidates. Build geographic monitoring of AI employment laws into your compliance calendar for 2026 and beyond.