The phrase “state AI employment laws” hides distinctions that determine what an organization must do now. A regulation already in force is not equivalent to a bill waiting in committee. Guidance may explain how an agency reads an existing civil-rights statute without creating a new AI code. A city ordinance may impose detailed duties inside city limits without governing the rest of the state.

This guide draws on Bot Labor Law’s nine verified state-level coverage areas as of September 8, 2026, along with tracked New York City municipal requirements. It is a deliberately scoped comparison, not a claim that these are the only federal, state, or local rules that could apply to workplace technology.

The current status snapshot

JurisdictionMeasureCategoryCurrent statusEmployment-AI issue
CaliforniaEmployment Regulations Regarding Automated-Decision SystemsState regulationEffective October 1, 2025Discrimination, records, disability-related inquiries
ColoradoSB 26-189 Automated Decision-Making Technology ActState statuteEnacted; core framework begins January 1, 2027Consequential employment decisions, notice, data rights, human review
IllinoisPublic Act 103-0804State statuteEffective January 1, 2026Discriminatory effects and employee notice
MarylandLabor and Employment § 3-717State statuteEffective October 1, 2020Consent for facial-template creation during job interviews
MassachusettsAttorney General AI advisoryAgency guidanceCurrent guidance on existing lawAnti-discrimination and other existing-law duties
New JerseyDCR algorithmic-discrimination guidanceAgency guidanceCurrent guidance on the existing LADAlgorithmic discrimination across employment decisions
New YorkState Technology Law article 4, as amendedState statuteEffective; public-sector scopeState and covered public-entity disclosures; employee rights
New York CityLocal Law 144 and implementing ruleMunicipal ordinance and ruleEffective; enforcement began July 5, 2023Bias audits, public summaries, candidate and employee notices
VermontH.340Pending state billIn House committee; not enactedProposed assessments, audits, disclosures, and governance
WashingtonHB 1672Pending state billRetained in House Appropriations in 2026; not enactedProposed monitoring limits, worker data rights, and human review

What is actually law now?

California’s Civil Rights Council regulations took effect October 1, 2025. They place automated-decision systems inside the state’s existing Fair Employment and Housing Act framework: covered use or selection criteria may violate that framework when it discriminates on a protected basis. The regulations also include automated-decision-system data in covered employment records generally retained for at least four years and address automated assessments that may elicit disability-related information.

Illinois took a direct statutory route. Public Act 103-0804 amended the Employment Article of the Illinois Human Rights Act and became effective January 1, 2026. It addresses AI used in recruitment, hiring, promotion, employment renewal, training or apprenticeship selection, discharge, discipline, tenure, and other employment terms. The statute prohibits covered AI use that has the effect of subjecting employees to protected-class discrimination, bars using zip codes as a proxy for protected classes in those activities, and requires notice when an employer uses AI for the specified employment purposes. The Illinois Department of Human Rights is directed to adopt implementation and enforcement rules.

Maryland’s law is narrower and more technology-specific. Labor and Employment § 3-717 prohibits an employer from using a facial-recognition service to create a facial template during an applicant’s job interview unless the applicant consents through a signed waiver containing the information specified by the statute. The provision has been effective since October 1, 2020. It should not be generalized into a statewide code for every biometric or automated hiring practice.

New York State’s operative record is also bounded. State Technology Law article 4 requires specified disclosures by state agencies. Chapter 86 of 2026 added, to the extent feasible, disclosures by defined covered entities that include specified local governments, school entities, SUNY, CUNY, and community colleges, and revised related public-employee protections. This is a public-sector framework, not a general private-employer AI hiring statute.

Colorado is enacted—but the compliance horizon is 2027

Colorado’s SB 26-189 became law on May 14, 2026. The act repealed and reenacted the state’s automated-decision framework and covers technology used to materially influence consequential decisions, including decisions about access to, eligibility for, or compensation related to employment. The official legislative summary places the core developer and deployer framework on a January 1, 2027 start date.

The enacted framework assigns technical-documentation duties to developers and notice and response duties to deployers. It addresses access to relevant personal data, correction of factually inaccurate data, and meaningful human review and reconsideration following specified adverse outcomes. The Attorney General has rulemaking and exclusive enforcement responsibilities under the Colorado Consumer Protection Act; the act does not create a new private right of action.

What remains pending?

Vermont H.340 remains a proposal. The official legislature lists its last recorded action as a February 25, 2025 first reading and referral to the House Committee on Commerce and Economic Development; the status page shows no enacted act. As introduced, H.340 would regulate developers and deployers of automated decision systems used for consequential decisions, including employment, through proposed assessment, audit, disclosure, notice, and governance provisions. Those provisions are not current Vermont requirements.

Washington HB 1672 is likewise pending. The House policy committee recommended a substitute and the bill was referred to House Appropriations on February 21, 2025. On January 12, 2026, it was reintroduced by resolution and retained in its existing status. The substitute would address electronic monitoring, automated employment decisions, worker data, and forms of human review. It has not passed the House and none of those proposed duties is operative law.

New York’s Boundaries on Technology Act, S 10147, offers another example of a private-sector proposal that should not be confused with the state-government law already in effect. Introduced April 30, 2026 and currently listed in the Senate Labor Committee, it would regulate electronic monitoring and automated employment decision tools through proposed impact assessments, notices, data rights, restrictions, and enforcement provisions.

California SB 7 belongs in the historical column, not the pending one. The 2025 proposal would have addressed notice, access, correction, appeal, and human oversight, but Governor Gavin Newsom vetoed it on October 13, 2025. Its proposed duties never became law.

Where guidance matters without becoming new law

The Massachusetts Attorney General’s April 16, 2024 advisory explains how existing consumer-protection, anti-discrimination, and data-security laws apply to AI developers, suppliers, and users. Its employment significance comes from the application of existing Massachusetts anti-discrimination law to AI-assisted practices. The advisory is not a newly enacted AI employment statute or regulation.

The New Jersey Division on Civil Rights took a similarly interpretive approach in January 2025. Its guidance explains that the existing Law Against Discrimination applies to algorithmic discrimination, including discrimination arising from tools used in employment advertising, screening, hiring, evaluation, promotion, discipline, and termination. The document expressly says it does not impose new or additional requirements, establish rights or obligations, or substitute for enforcement of the LAD.

The practical point is not that guidance is irrelevant. It states how the enforcing agency understands existing authority and may clarify risk in vendor selection, tool design, and deployment. But its legal category must remain visible: the underlying statutes supply the requirements; the guidance explains the agency’s interpretation.

Why New York City matters

New York City Local Law 144 demonstrates how municipal rules can be more prescriptive than the statewide private-employer baseline represented in this dataset. The city ordinance regulates qualifying automated employment decision tools used to screen candidates for employment or employees for promotion within its defined city scope. It requires a recent bias audit, public availability of specified audit information, and notices to covered candidates and employees. The ordinance took effect January 1, 2023; the initial implementing rule became effective May 6, 2023; an updated rule and DCWP enforcement followed on July 5, 2023.

Those duties do not apply merely because software uses AI. Coverage depends on the ordinance and rule definitions, the tool’s role in the decision, and the New York City nexus. The measure is significant precisely because it is specific: a local audit-and-notice regime for defined hiring and promotion tools, not a statewide ban on automated employment systems.

The patterns are functional, not uniform

  • Existing civil-rights law remains a central legal anchor. California and Illinois expressly connect automated employment practices to discrimination rules, while Massachusetts and New Jersey explain the application of existing anti-discrimination law through guidance.
  • Some measures regulate a narrow practice. Maryland focuses on facial-template creation during interviews; New York City focuses on qualifying tools used for hiring and promotion screening.
  • Broader frameworks allocate responsibility across a system lifecycle. Colorado separates developer documentation from deployer notices and response duties, while its definitions and exemptions determine coverage.
  • Transparency takes different forms. The dataset includes public audit summaries, employee or candidate notices, state-agency disclosures, technical documentation, and post-outcome explanations—not one standardized disclosure model.
  • Human review appears most clearly in future-effective or proposed frameworks. Colorado includes meaningful human review following specified adverse outcomes; pending bills in Vermont, Washington, and New York propose different forms of oversight or reconsideration.
  • Municipal authority can create the most operationally detailed rule in a jurisdiction. New York City’s requirements must remain distinct from New York State law and pending statewide bills.

What to watch next

  1. Colorado implementation before January 1, 2027, including Attorney General rulemaking on required disclosures and the practical division of developer and deployer responsibilities.
  2. Illinois Department of Human Rights rules addressing when, how, and under what conditions the statutory employee notice must be provided.
  3. Any official action on Vermont H.340, Washington HB 1672, or New York S 10147. Until a legislature and governor complete enactment, their proposed obligations remain proposals.
  4. Agency updates or enforcement materials interpreting existing law in Massachusetts and New Jersey.
  5. New York City enforcement and guidance on the defined scope, audit process, public summary, and notice mechanics of Local Law 144.

Start with status, not the headline

A useful state AI employment-law review begins with four questions: What kind of authority is this? Has it been enacted? When do the relevant provisions apply? Which employers, workers, decisions, and technologies fall within its scope? Those questions separate current compliance work from legislative monitoring and agency interpretation.

Bot Labor Law’s tracker preserves those distinctions at the record level, with separate dates, jurisdiction labels, histories, and official sources. Use it to check the underlying measure before treating any summary—including this guide—as the final word.