AI Bias Audit Laws by State 2026
There is no single federal AI hiring law — so compliance is a 50-state patchwork, and it's tightening fast. If you build HR tech or use AI to screen candidates, where your applicants are located now determines what you owe. Here's the state-by-state map as it stands in 2026.
Why This Is a Patchwork (and Why It Matters)
Federal anti-discrimination law (Title VII, the ADA, the ADEA) already prohibits discriminatory outcomes, and the EEOC has signaled that AI hiring tools are squarely within its enforcement reach. But there is no federal statute that specifically mandates AI bias audits. That vacuum is being filled state by state — each with its own definitions, triggers, and deadlines.
For HR tech vendors, this means a single product can face multiple, overlapping obligations depending on where customers operate. For employers, it means the location of the role — not just your HQ — drives which rules apply. Below is the current landscape.
State-by-State Breakdown
New York City — Local Law 144
IN EFFECT — MANDATORY AUDITThe most prescriptive law in the country. Employers and employment agencies using an Automated Employment Decision Tool (AEDT) to screen NYC candidates/employees must: (1) have an independent bias audit completed within the prior year, (2) publish a summary of audit results on their website, and (3) notify candidates at least 10 business days before use. Enforcement is active.
Action: If you touch NYC roles, get an independent bias audit and publish the summary now.
Colorado — AI Act (SB 24-205)
DUTY OF REASONABLE CAREThe first comprehensive US state AI law. Developers and deployers of high-risk AI used in consequential decisions — explicitly including employment — owe a duty of reasonable care to protect consumers from algorithmic discrimination. Requires impact assessments, consumer disclosures, and notice when AI is used in a decision. Effective date has been subject to legislative adjustment — verify the current timeline.
Action: Prepare impact assessments and disclosure workflows; confirm the in-force date with counsel.
Illinois — AIVIA + Human Rights Act
VIDEO + DISCRIMINATION RULESIllinois' Artificial Intelligence Video Interview Act requires notice, consent, and explanation when AI analyzes video interviews, plus demographic reporting in some cases. A 2026 amendment to the Illinois Human Rights Act addresses AI use in employment decisions and makes discriminatory use of AI an unlawful practice, with notice obligations.
Action: If you use AI on video interviews or hiring decisions for IL roles, add notice/consent flows.
California — ADMT & Civil Rights Rules
ADMT + DISCRIMINATIONCalifornia is regulating from two directions: the CPPA's automated decision-making technology (ADMT) regulations add access and opt-out style rights for significant decisions including employment, and the Civil Rights Council adopted rules clarifying that using AI/automated-decision systems that cause discriminatory impact violates state anti-discrimination law. Confirm current effective dates for both tracks.
Action: Map any AI used in CA hiring; prepare for notice, access, and anti-discrimination scrutiny.
Other States — Pending / Emerging
WATCH LISTMultiple states (e.g., New Jersey guidance, and bills introduced in states such as New York statewide, Massachusetts, Washington, and others) have proposed or issued guidance on AI in employment. Texas enacted a broader AI governance law (TRAIGA) touching government and certain uses. The landscape is shifting every legislative session.
Action: Maintain a monitoring process; treat new-state expansion as a recurring compliance task.
What "Covered" Actually Means
The hardest part of these laws is figuring out whether a given tool is in scope. Common triggers across the statutes:
- The tool substantially assists or replaces human judgment — NYC LL144's AEDT definition turns on whether the tool's output is weighted heavily or overrides human decision-making, not whether it's the sole decider.
- It's used for a consequential employment decision — hiring, screening, promotion, termination, compensation. Scheduling or purely administrative automation is usually outside scope.
- The candidate or role is located in the regulating jurisdiction — geography of the applicant/position, not the employer's HQ, is typically the hook.
Compliance Obligations Compared
Independent Bias Audit
NYC LL144 mandates an annual audit by an independent auditor with a published summary. Other states lean on impact assessments rather than a formal public 'audit', but documenting disparate-impact testing is best practice everywhere.
Candidate Notice
NYC, Illinois, and Colorado all require some form of advance notice that AI is being used in the process. Build notice into your application flow; retrofit is painful across multiple jurisdictions.
Impact Assessments
Colorado and California push toward documented impact/risk assessments for high-risk or significant-decision AI. These resemble EU AI Act and GDPR DPIAs — reuse the same internal process where possible.
Recordkeeping & Disclosure
Across the board, regulators expect you to be able to show what tool was used, how it was evaluated for bias, and what you told candidates. Maintain an auditable trail per tool, per jurisdiction.
Multi-State AI Hiring Compliance Checklist
For HR tech vendors and employers operating across states. Steps 1–4 are the highest-leverage.
This article is general information, not legal advice. State AI employment laws, effective dates, and penalty figures change frequently — Colorado and California in particular have revisited timelines. Verify current requirements with employment counsel before acting.
Keep your hiring tools and careers pages compliant
Bias-audit notices and AI disclosures only work if every candidate can actually access them. RatedWithAI scans your careers pages and application flows for accessibility and compliance gaps. Start with a free scan.
Scan Your Hiring Pages for Free →Frequently Asked Questions
Which states regulate AI bias in hiring?
The most significant 2026 regimes are New York City (Local Law 144's mandatory bias audits for automated employment decision tools), Colorado (the AI Act's duty of reasonable care for high-risk AI in consequential decisions, including employment), Illinois (the AI Video Interview Act plus a 2026 Human Rights Act amendment on AI discrimination), and California (CPRA ADMT rules and Civil Rights Council guidance). Other states have pending bills and guidance.
What does an AI bias audit involve?
It's an impartial evaluation of an automated decision tool for disparate impact — comparing selection or scoring rates across protected groups such as sex, race/ethnicity, and intersectional categories. NYC Local Law 144 requires the audit be performed by an independent auditor within the prior year, with a public summary of the results posted on the employer's website.
Does NYC Local Law 144 apply if my company isn't in New York?
It can. The law keys off the location of the job, not the employer's headquarters. If you use an automated employment decision tool to screen candidates or employees for positions located in New York City, you're covered — even as a remote-first company based elsewhere. You must run the bias audit, publish the summary, and provide candidate notice.
Is there a federal AI hiring law?
There is no federal statute specifically mandating AI bias audits, but existing federal anti-discrimination laws (Title VII, ADA, ADEA) already apply to AI-driven hiring outcomes, and the EEOC has stated that algorithmic tools causing disparate impact can violate those laws. So even outside the regulated states, discriminatory AI hiring carries federal legal exposure.
What happens if I ignore these laws?
Penalties vary by jurisdiction. NYC LL144 imposes civil penalties per violation per day. Colorado and California tie enforcement to their attorneys general and agencies, with potential injunctive relief and fines. Beyond statutory penalties, a documented failure to test for bias strengthens private discrimination claims and class actions, which are often the bigger financial risk.