Every US jurisdiction that requires independent bias audits or impact assessments for AI systems used in employment, housing, and other high-risk decisions.
AI bias audits are becoming a baseline compliance requirement for companies deploying automated decision tools in hiring, lending, and insurance. Below are the US jurisdictions with enacted laws or finalized regulations that mandate bias audits or algorithmic impact assessments. Each entry links to the official primary source and, where available, our detailed analysis.
| Jurisdiction | Law / Bill # | What Requires an Audit | Audit Frequency | Deadline | Penalty | Source |
|---|---|---|---|---|---|---|
| New York City | Local Law 144 | Automated employment decision tools (AEDTs) used to screen or evaluate candidates for employment or promotion | Annual independent bias audit | Enacted Jan 1, 2023; DCWP enforcement effective July 5, 2023 | $500 first violation; $500–$1,500 each subsequent violation per day | Official source |
| Colorado | SB 26-189 (ADMT law; repeals and replaces SB 24-205 on Jan 1, 2027) | Automated decision-making technology that materially influences consequential decisions (education, employment, housing, financial/lending, insurance, health care, essential government services) | No audit or impact-assessment mandate — SB 26-189 drops SB 24-205's impact assessments (which apply until Jan 1, 2027); requires pre-use notice, adverse-outcome explanations and 3-year record retention | Applies to consequential decisions made on or after January 1, 2027 (SB 24-205's duties, including impact assessments, have applied since June 30, 2026) | Enforced exclusively by the Colorado AG under the Colorado Consumer Protection Act; no private right of action; 60-day right to cure (sunsets Jan 1, 2030) | Official source |
| Illinois | AI Video Interview Act (820 ILCS 42) | AI analysis of video interviews for employment; requires notice, consent, and data handling obligations | Consent required for each use; no fixed audit cadence | Effective January 1, 2020 (reporting section added effective January 1, 2022) | The Act contains no enforcement, penalty, or private-right-of-action section; it is not part of the Illinois Human Rights Act | Official source |
| Illinois | HB 3773 (Human Rights Act amendment) | AI used in employment decisions including hiring, promotion, discipline, and termination | Notice to applicants/employees; no impact-assessment requirement | Effective January 1, 2026 (Public Act 103-0804) | Enforced by Illinois Department of Human Rights; civil penalties under IHRA | Official source |
| California | FEHA Automated Decision Regulations (CRD) | Automated decision systems used in employment decisions (hiring, promotion, termination) | No fixed audit cadence; the regulations apply existing anti-discrimination law to automated-decision systems | Finalized June 2025; effective October 1, 2025 (California Civil Rights Department) | FEHA enforcement: CRD investigation, civil penalties, damages | Official source |
Sources: Official state legislature websites, NYC DCWP, Colorado General Assembly, Illinois General Assembly, California CRD. Last updated September 2026.
Describe your AI system and we'll connect you with compliance specialists who understand your jurisdiction's requirements.
Get Compliance HelpFree consultation request · No obligation
An AI bias audit is an independent evaluation of an automated decision system to assess whether it produces discriminatory outcomes based on protected characteristics such as race, gender, or age. NYC Local Law 144, for example, requires an "independent auditor" to test automated employment decision tools (AEDTs) for disparate impact by sex and race/ethnicity. Colorado's SB 24-205 requires broader "impact assessments" for high-risk AI systems until January 1, 2027, when SB 26-189 replaces it and requires notices and adverse-outcome explanations instead. See our AI hiring laws guide for a state-by-state breakdown.
Any employer or employment agency using automated tools to screen, evaluate, or select job candidates in New York City must conduct annual bias audits under Local Law 144. In Colorado, SB 24-205 requires deployer impact assessments until January 1, 2027; SB 26-189, which replaces it on that date, does not require audits or impact assessments, and instead requires notices and adverse-outcome explanations when automated decision-making technology materially influences consequential decisions. Illinois requires notice and consent for AI video interviews under AIVII (820 ILCS 42), and HB 3773 (effective January 1, 2026) bars discriminatory AI use in employment decisions and requires notice.
Audit frequency varies by jurisdiction. NYC Local Law 144 requires audits at least annually, with results published on the employer's website. Colorado SB 26-189 sets no audit cadence (SB 24-205's annual impact assessments end when SB 26-189 replaces it on January 1, 2027). Illinois HB 3773 requires notice to applicants and employees but does not specify an audit cadence. California's Civil Rights Council regulations (effective October 1, 2025) clarify how existing anti-discrimination law applies to automated-decision systems in employment.
Documentation requirements include: (1) a description of the AI system's purpose and intended use, (2) the data used for training and testing, (3) the metrics evaluated (e.g., selection rates by demographic group), (4) the results of statistical tests for disparate impact, (5) any mitigation steps taken, and (6) a summary of findings published for public review (required by NYC LL 144). Colorado SB 26-189 requires developers to give deployers documentation on intended uses, known risks and limitations, and training-data categories, and requires records to be kept for at least three years.
Penalties vary: NYC imposes $500 for a first violation and $500–$1,500 for each subsequent violation per day under Local Law 144, enforced by the Department of Consumer and Worker Protection. Colorado's SB 26-189 (from January 1, 2027) is enforced exclusively by the Attorney General under the Colorado Consumer Protection Act, with no private right of action. The Illinois AI Video Interview Act has no enforcement or penalty section; Illinois Department of Human Rights enforcement applies to HB 3773 under the Human Rights Act. California FEHA violations can result in enforcement actions by the Civil Rights Department.
NYC Local Law 144 explicitly requires an "independent auditor" — someone who is not involved in the development or use of the AEDT being audited. Colorado SB 26-189 does not require audits or impact assessments. Illinois HB 3773 does not require impact assessments or audits. Best practice across all jurisdictions is to use a qualified third-party auditor to ensure objectivity.
The EU AI Act (Regulation (EU) 2024/1689) requires conformity assessments for high-risk AI systems before they can be placed on the EU market, including bias testing under Article 10 (data governance) and Article 9 (risk management). US state laws are narrower in scope: NYC covers only employment AEDTs, Colorado (SB 26-189, from 2027) covers automated consequential decisions across multiple domains, and Illinois focuses on video interviews and hiring. See our EU AI Act vs US State Laws crosswalk for a detailed side-by-side comparison.