The Growing Patchwork Of State AI Laws: What It Means For Employers

Key Takeaways

• Connecticut’s Artificial Intelligence Responsibility and Transparency Act establishes a comprehensive framework for automated employment decision technology, with core obligations phasing in between October 1, 2026, and October 1, 2027.
• The law amends the Connecticut Fair Employment Practices Act to make clear that the use of automated employment decision technology is not a defense to discrimination claims, while permitting courts and the CHRO to consider an employer’s anti-bias testing efforts as a mitigator.
• Connecticut joins a growing patchwork of state AI laws—including measures in Illinois, Colorado, California, New York City and Texas—that differ in enforcement, scope and notice and audit obligations, creating a complex compliance challenge for multi-state employers.
• With the patchwork of state requirements coupled with the absence of federal statutes specifically addressing the use of AI in employment, employers should consider whether to inventory AI tools across the employment lifecycle, conduct bias audits, update vendor contracts, and/or revise WARN Act protocols.

Overview

On May 1, 2026, the Connecticut General Assembly passed Senate Bill 5, the Artificial Intelligence Responsibility and Transparency Act (SB 5 or the Act). Governor Ned Lamont has indicated that he will sign the bill into law. Once enacted, Connecticut will become one of the U.S. jurisdictions to comprehensively regulate the use of AI that impacts employment, joining Illinois, Colorado, California, New York City and Texas in a rapidly developing—and increasingly inconsistent—state-level regulatory landscape.

This alert summarizes the key provisions of SB 5, places it in the broader context of state AI employment regulation, surveys the federal landscape, and offers practical guidance for employers navigating the emerging compliance environment.

Connecticut’s Artificial Intelligence Responsibility and Transparency Act (SB 5)

Most relevant for employers, SB 5 imposes obligations on developers and deployers of “automated employment-related decision processes” (AERDP), amends the Connecticut Fair Employment Practices Act, and expands disclosure requirements under the state’s Worker Adjustment and Retraining Notification (WARN) Act. The Act takes effect in stages.

Notable Dates

• October 1, 2026: The automated employment decision technology framework takes effect; “AI is not a defense” amendments to the state’s anti-discrimination statutes apply; the WARN Act AI disclosure requirement begins; and frontier developer whistleblower provisions take effect.
• October 1, 2027: Interactive disclosure and pre-decision notice obligations for automated employment decision technology take effect.
• December 31, 2027: Before initiating enforcement, the Attorney General must afford a 60-day period to cure for violations alleged on or before December 31, 2027.

1. Automated Employment Decision Technology

SB 5 broadly defines AERDP to include any technology that processes personal data and produces outputs (e.g., scores, recommendations or classifications) that are a substantial factor in an employment decision. As such, AERDP may include resumé screening tools, candidate ranking algorithms, performance management software and automated promotion engines.

Under the law, developers of AERDP must provide deployers (i.e., employers) with the compliance information necessary to satisfy the Act’s notice and disclosure obligations. Deployers, in turn, must:

• Notify employees and applicants, in plain language, when AERDP is used in connection with an employment decision.
• Provide a pre-decision notice that identifies the purpose of the tool, its trade name, the categories and sources of data used, how that data is assessed and contact information for further inquiries.
• Make available an interactive disclosure mechanism allowing applicants and employees to seek additional information.

The Act includes a trade secret safe harbor that allows developers and deployers to withhold protected information, but they must disclose what is being withheld and specify the legal grounds for the withholding.

2. “AI Is Not a Defense”

SB 5 amends the Connecticut Fair Employment Practices Act to clarify that the use of automated employment decision technology is not a defense to a discrimination claim. An employer cannot avoid liability simply by attributing an adverse outcome to a vendor’s algorithm.

Importantly, however, courts and the Commission on Human Rights and Opportunities (CHRO) may consider evidence of an employer’s anti-bias testing and proactive compliance efforts as a mitigating factor. Relevant factors include the quality, efficacy, recency and scope of any testing; the results obtained; and the employer’s response to those results. Robust, well-documented governance can therefore meaningfully shape exposure even though it is not a complete defense.

3. WARN Act Disclosure

Beginning October 1, 2026, employers filing notices under Connecticut’s WARN Act must disclose whether the layoffs are related to AI or other technological change. This disclosure obligation will require HR, legal and operations teams to coordinate closely on the rationale for workforce reductions and to maintain contemporaneous documentation supporting the stated cause.

4. Enforcement

Violations of SB 5 are deemed unfair or deceptive trade practices under the Connecticut Unfair Trade Practices Act (CUTPA). Enforcement authority lies exclusively with the Connecticut Attorney General; the Act creates no private right of action. Through December 31, 2027, the Attorney General must afford alleged violators a 60-day period to cure before initiating enforcement.

The Broader State AI Employment Landscape

Connecticut’s Act takes shape against a backdrop of significant—and increasingly divergent—state activity. The most consequential measures for employers are summarized below.

California

California’s regulatory landscape is layered. Amendments to the Fair Employment and Housing Act (FEHA) (Cal. Gov. Code § 12940) addressing automated decision systems (ADS) became effective October 1, 2025; they prohibit ADS that discriminate based on protected characteristics and require four-year retention of ADS-related data. California’s FEHA regulations expressly incorporate the federal Uniform Guidelines on Employee Selection Procedures (29 C.F.R. Part 1607; Cal. Code Regs. tit. 2, § 11017), requiring that any “testing device or other means of selection” with an adverse impact be job-related and consistent with business necessity. Separately, the California Privacy Protection Agency’s final CCPA regulations took effect January 1, 2026, with risk assessment obligations beginning then for covered businesses; automated decision-making technology (ADMT) requirements for significant decisions begin January 1, 2027, including pre-use notices, explanations of how the tool works, and opt-out rights. Covered business comprise those with $25 million or more in revenue or that handle significant volumes of consumer data.

Colorado (SB 26-189)

Colorado’s Senate Bill 26-189 (S.B. 26-189, signed into law May 14, 2026, anticipated codification at Colo. Rev. Stat. §§ 6-1-1701 et seq.), which repeals and replaces the state’s original 2024 AI Act (SB 24-205), takes effect January 1, 2027, and applies to ADMT used to materially influence consequential decisions in areas including employment (e.g., hiring, promotion, discipline or discharge), education, housing, financial services, health care and insurance. Deployers must provide pre-use notices, make post-adverse outcome disclosures within 30 days, maintain three-year records and provide consumers experiencing adverse outcomes with meaningful human review and rights to correct inaccurate personal data. Developers must provide deployers with technical documentation describing intended uses, training data categories, known limitations and instructions for appropriate use and human review. Violations are enforceable by the Attorney General as deceptive trade practices under the Colorado Consumer Protection Act. The Attorney General has exclusive enforcement authority with a 60-day cure period for non-knowing violations, and there is no private right of action.

Illinois (HB 3773)

Effective January 1, 2026, Illinois HB 3773 (codified at 775 ILCS 5/2-101 & 5/2-102) prohibits employers from using AI that results in discrimination based on protected characteristics, including unintentional disparate impact. Employers must provide notice to applicants and employees when AI is used in employment decisions and may not use ZIP codes as a proxy for protected characteristics. Draft regulations released by the Illinois Department of Human Rights specify the required content of notices and impose a four-year recordkeeping requirement. There is no private right of action; complaints are filed with the Illinois Department of Human Rights.

New York City (Local Law 144)

In effect since July 5, 2023, New York City’s Local Law 144 (N.Y.C. Admin. Code §§ 20-870 to 20-874) prohibits employers and employment agencies from using an “automated employment decision tool” (AEDT), defined as an automated process or AI that is used to substantially assist or replace employment discretionary decision making, unless certain conditions are met. Employers must conduct annual independent bias audits, publicly post audit summaries, and provide candidates with at least ten business days’ notice before using an AEDT. Fines range from $500 to $1,500 per day for violations. The law applies even when humans make final decisions based on AI-generated rankings or scores.

Texas (TRAIGA)

The Texas Responsible Artificial Intelligence Governance Act (TRAIGA) (Tex. Bus. & Com. Code §§ 552.001 et seq.), effective January 1, 2026, is a broad AI governance statute with consumer-facing transparency, prohibited-practices provisions and exclusive enforcement by the Texas Attorney General. It prohibits only intentional AI-based discrimination and does not extend to disparate impact liability, imposes no audit or notice obligations and provides a 60-day notice and cure period. The Texas approach reflects a materially lighter compliance burden than that adopted in Connecticut, Illinois, Colorado or California.

The Federal Landscape

There is currently no federal statute specifically governing the use of AI in employment. Title VII of the Civil Rights Act of 1964 (42 U.S.C. § 2000e-2), the Americans with Disabilities Act (42 U.S.C. §§ 12101 et seq.) and the Age Discrimination in Employment Act (29 U.S.C. §§ 621 et seq.) apply with equal force regardless of whether an employment decision is made by a human or an algorithm. Under these statutes, employers may be liable for both disparate treatment (intentional discrimination) and disparate impact (facially neutral practices that disproportionately affect protected groups). Use of a third-party vendor’s AI tool does not insulate employers from liability; the Uniform Guidelines on Employee Selection Procedures make clear that “[t]he use of an employment agency does not relieve an employer…of its responsibilities under Federal law…” (29 C.F.R. § 1607.10). The Equal Employment Opportunity Commission has identified technology-related employment discrimination as an enforcement priority in its Strategic Enforcement Plan for fiscal year (FY) 2023–2027 and has issued guidance confirming that its enforcement authority extends to AI-driven hiring tools.

Recent federal activity has focused largely on rolling back state AI regulation. A December 2025 executive order directed the Department of Justice to challenge “burdensome” state AI laws, and the March 2026 National AI Legislative Framework recommended federal preemption of state employment-related AI requirements. As of May 2026, the White House is circulating a draft executive order calling for federal review of advanced AI models in the wake of the Anthropic Mythos release. Despite these federal initiatives, no preemption legislation has been enacted, and existing state laws remain operative—meaning employers must continue to plan for state compliance.

Practical Guidance for Employers

Given the patchwork of state requirements, employers—particularly those operating in multiple jurisdictions—should consider whether the following might be helpful:

1. Inventorying all AI tools used across the employment lifecycle.
2. Conducting bias audits and impact assessments.
3. Implementing notice, consent and accommodation protocols for candidates and employees.
4. Renegotiating vendor contracts.

Looking Ahead

Connecticut’s SB 5 is unlikely to be the last word. Additional states are expected to introduce comparable legislation in the coming sessions, and the contours of federal preemption efforts will continue to evolve. Employers that build flexible governance frameworks now—grounded in inventory, audit, notice and vendor management—will be well positioned to implement new requirements as they emerge.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.