AI Hiring Laws by US State in 2026: Every Active AEDT Regulation

New York City Local Law 144: the AEDT framework everyone copied

NYC Local Law 144 has been in force since July 5, 2023 and remains the most operationally specific AI hiring law in the US. It applies to any employer or employment agency that uses an automated employment decision tool to substantially assist a hiring or promotion decision for a position located in NYC. The rule defines an AEDT as any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence that issues simplified output (a score, classification, or recommendation) and substantially assists the decision. The five word phrase that triggers the law is automated employment decision tool, and the Department of Consumer and Worker Protection has been clear that it reads substantially assists broadly.

Compliance turns on three pillars. First, an independent bias auditor must run a written audit within one year of tool use and the summary results must be posted on the careers page. Second, candidates must get notice at least 10 business days before the AEDT is used, including the job qualifications the tool assesses and how to request an alternative selection process. Third, an accommodation must be available on request. Penalties run from $500 for a first violation up to $1,500 per subsequent violation, and each day a tool is in use without a current audit is a separate violation. The first DCWP enforcement actions landed in 2024.

Illinois HB 3773: the broadest state ban on AI discrimination in employment

Illinois HB 3773 amends the Illinois Human Rights Act and took effect on January 1, 2026. The amendment makes it a civil rights violation for an employer to use AI in a way that has the effect of subjecting employees to discrimination across the 12 protected categories of the IHRA. The list is broader than Title VII and includes military status, sexual orientation, and pregnancy, so the Illinois rule reaches conduct that federal law may not reach directly. Employers must give written notice when AI is used in covered employment decisions including recruitment, hiring, promotion, discharge, discipline, or terms of employment.

The Illinois Department of Human Rights investigates complaints, issues findings, and can impose civil penalties up to $50,000 per violation in addition to the standard IHRA remedies of back pay, reinstatement, and damages. The statute does not require a formal bias audit, but the practical reality is that you need adverse impact data to defend an investigation. The Illinois Department of Labor has signaled that it will read the notice duty strictly, so generic AI disclaimers in candidate handbooks are not enough.

Colorado SB 24-205: the first comprehensive risk management law for hiring AI

Colorado SB 24-205, called the Colorado AI Act, took effect on February 1, 2026. It is the first US state law that imposes comprehensive risk management duties on developers and deployers of high risk AI systems. Employment AI used in hiring, promotion, discipline, or termination is high risk by definition under the statute. Deployers must complete impact assessments, notify candidates that AI is in use, give a meaningful explanation of how the system works, and provide an opportunity to appeal an adverse decision.

The Colorado Attorney General enforces, with civil penalties scaled to the violation. The statute gives a safe harbor to deployers that maintain a written risk management program based on a recognized framework such as the NIST AI Risk Management Framework. Practically, that means documented model cards, vendor due diligence, validation data, and post deployment monitoring. The Colorado AI Act overlaps with Illinois HB 3773 in scope, so multi state employers should align programs to the higher standard.

California CRC ADS regulations: FEHA, extended to automated decision systems

The California Civil Rights Council finalized regulations in 2026 that bring automated decision systems under the Fair Employment and Housing Act. The rules apply to any employer with 5 or more employees in California, plus covered agents and vendors that operate AI tools for employment decisions on behalf of an employer. The core principle is straightforward: an automated decision system that produces disparate impact violates FEHA the same way a biased interview question would. The California Civil Rights Department enforces, with standard FEHA remedies including back pay, reinstatement, and damages.

Recordkeeping duties extend to ADS inputs and outputs. Employers must keep at least the prior four years of decision data showing how the tool scored candidates, what inputs were used, and the demographic breakdown of selection rates. Vendors are now exposed as covered agents under the regulation, which is a meaningful shift from prior California practice where the employer alone carried the risk.

Maryland HB 1202: written consent for facial recognition in interviews

Maryland HB 1202 took effect on October 1, 2020 and was clarified in 2024. The law requires written consent from an applicant before an employer uses facial recognition technology during a job interview. The 2024 guidance from the Maryland Commission on Civil Rights confirmed that AI driven facial analysis and emotion detection used in video interview platforms (the modern HireVue category) fall inside the consent rule.

Without a signed waiver, an employer cannot lawfully analyze a candidate face during an interview, whether the video is live or recorded. The Maryland Commission on Civil Rights enforces, and there is a private right of action so a candidate can sue directly. The practical impact: any video interview vendor used in Maryland needs an opt in consent workflow at the start of the interview, not buried in a privacy policy.

Federal EEOC overlay: Title VII, the ADA, the ADEA, and OFCCP

Federal law applies on top of every state and city rule above. The EEOC issued technical guidance in 2022 on the ADA and AI selection tools and follow on guidance in 2023 on Title VII and AI. Both were reaffirmed in 2025. The guidance makes clear that an employer is responsible for any selection procedure including a third party AI tool, and that the four-fifths rule from 29 CFR Part 1607 remains the baseline test for adverse impact under Title VII. The Age Discrimination in Employment Act (ADEA) reaches AI tools that screen out older workers, and the ADA reaches tools that screen out people with disabilities.

OFCCP adds federal contractor obligations on top. Federal contractors and subcontractors must keep records of selection procedures, including AI inputs and outputs, and must comply with the affirmative action plan rules in 41 CFR Part 60. Debarment is a real risk for contractors who fail to validate or document their AI selection tools. Treat federal compliance as the floor, not the ceiling.

Practical compliance checklist for HR using AI in hiring

The list below is the working checklist our editorial team uses when an HR or talent acquisition leader asks how to deploy an AI hiring tool without picking up a regulator letter. It is not legal advice and it is not exhaustive, but it is the floor. Verify against the law in every state where you hire and run a counsel review before deployment.

• Identify every automated employment decision tool in use. Inventory resume screeners, video interview platforms, chatbot prescreens, skills tests scored by AI, and ranking or matching engines. NYC Local Law 144 turns on whether a tool substantially assists or replaces a hiring decision, so the inventory must include vendor tools your recruiters use.
• Commission an independent bias audit. NYC requires a written audit by an independent auditor within one year of use. The audit must compute selection rates across race, ethnicity, and sex, and produce a public summary. Illinois and Colorado push toward similar discipline through risk management duties.
• Run a four-fifths rule adverse impact analysis. The federal four-fifths rule from the Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607) remains the baseline test the EEOC uses. If the selection rate for any protected group is less than 80 percent of the rate for the top group, you have prima facie adverse impact and you need a job related defense.
• Post candidate notice at least 10 business days in advance. NYC Local Law 144 requires the notice on the careers page or in the job posting. Illinois HB 3773 and Colorado SB 24-205 also require written notice when AI is used. Include the tool name, the qualifications or characteristics it assesses, and how to request an accommodation.
• Document an accommodation path under the ADA. Candidates with disabilities have the right to request an alternative selection method. The EEOC ADA guidance on AI makes this explicit. Train recruiters to grant reasonable alternatives without delay and keep a written record of the request and the resolution.
• Get written consent for facial or voice analysis. Maryland HB 1202 requires written consent for facial recognition in interviews. Illinois BIPA already requires written consent for biometric collection, including voiceprints. Treat any AI that scores facial expressions or speech patterns as biometric for risk planning purposes.
• Keep records of inputs, outputs, and version history. Title VII and OFCCP recordkeeping rules require retention of selection data. The California ADS regulations and Colorado SB 24-205 extend this to AI inputs and outputs. Keep at least two years of model version, training data summary, prompts, and decision logs.
• Get counsel review before deployment. Every new AI hiring tool, every new state of operation, and every model version change should pass through employment counsel. State law moves faster than this page can. The disclaimer at the top of this page is not boilerplate.

Pending and at risk bills to watch

Several bills sit in committee or in active rulemaking as of May 2026. New Jersey S1588 and A3854 mirror the NYC AEDT framework with a bias audit duty and candidate notice; the bills are pending and there is no effective date yet. Washington has multiple AI in employment bills that have cleared committee but not the floor. Massachusetts HD 4029 reaches automated decision tools in employment but remains in draft form. Texas HB 149 (TRAIGA) took effect on January 1, 2026 with broad consequential decision scope that touches some employment uses, and the Texas Attorney General has issued early guidance. Track the National Conference of State Legislatures AI legislation tracker for weekly updates and the state legislature site for the bill text itself.

The verdict: how to think about US AI hiring law in 2026

My read after working through every active rule: NYC Local Law 144 is the operationally hardest, Illinois HB 3773 is the broadest, and Colorado SB 24-205 is the most strategic because it ties hiring AI into a risk management framework that the rest of the country will probably copy. If you are a multi state employer, build your program to Illinois plus Colorado plus NYC plus California, and you have covered roughly 35 percent of the US labor market and most of the demanding standards in one stack. Add Maryland written consent for any video interview product. Layer EEOC and OFCCP on top, and you have a defensible federal baseline. Verified May 2026.

The compliance lesson is that you cannot wait for federal law to catch up. The federal government has not passed a standalone AI hiring statute, but the EEOC has been explicit that Title VII and the ADA apply, and the state and city rules are real, in force, and enforced. The single highest leverage action for any HR or talent team in 2026 is to inventory every AI in the hiring stack, get an independent bias audit on each one, and put a written candidate notice on every job posting. The other items in the checklist follow from there.