Connecticut’s AI Rules Take Shape: What Employers and Developers Need to Know

On May 27, 2026, Connecticut Gov. Ned Lamont signed SB 5 (Public Act No. 26-15), a sweeping AI law with new obligations for employers that use AI in employment decisions and for the developers that build those tools. This alert summarizes SB 5’s key employment-related provisions and outlines practical next steps. It does not address SB 5’s other provisions, such as those regulating subscription-based AI providers, AI companion products and synthetic digital content.

What tools are covered?

Coverage under SB 5 turns on three definitions:

• Automated employment-related decision technology (AEDT): Any technology that processes personal data and uses computation to generate an output (e.g., a prediction, recommendation, ranking or score) that is a “substantial factor” used to make or materially influence an “employment-related decision.” Excluded are word processing, spreadsheet, database and similar software that does not make or materially influence such decisions; systems used only incidentally to a decision; and purely descriptive, diagnostic or statistical information not relied upon in a decision.
• Substantial factor: Any factor, including a constraint, ranking, score, recommendation or classification, that meaningfully alters the outcome of an employment-related decision concerning an individual in Connecticut.
• Employment-related decision: Any decision, based on an individual’s personal data, to hire, promote, discipline, discharge, renew employment, select for training or apprenticeship, or otherwise affect tenure or the terms, privileges or conditions of employment. Excluded are decisions resulting in nonmaterial changes to job tasks, responsibilities, hours, or assignments and decisions concerning workplace health and safety, scheduling and planning, or productivity monitoring.

Notably, the “materially influence” standard appears broader than New York City’s AEDT law, which covers only tools that “substantially assist or replace discretionary decision making,” and may reach tools that merely inform a human decision-maker.

Employers’ disclosure and notice obligations beginning October 1, 2027

These obligations apply to “deployers”: any person doing business in Connecticut that deploys an AEDT in the state. Unfortunately, SB 5 defines neither phrase, leaving unclear whether the obligations reach out-of-state employers with Connecticut-based remote workers or applicants, or Connecticut applicants for out-of-state roles. Until further clarity emerges, multistate employers may wish to take a conservative approach and assume coverage.

Disclosure of AEDT interactions

Employers must disclose in plain language to employees and applicants in Connecticut when they are interacting with an AEDT (e.g., an AI chatbot or automated interview platform), unless that would be obvious to a reasonable person.

Pre-decision written notice to employees and applicants

Before using an AEDT to make, or as a substantial factor in making, an employment-related decision about an employee or applicant in Connecticut, employers must provide that individual a written notice disclosing:

• That the employer has deployed an AEDT.
• The purpose of the AEDT and the nature of the employment-related decision.
• The trade name of the AEDT.
• The categories of the individual’s personal data the AEDT will analyze or process, and how that data will be assessed in reaching a decision.
• The sources of that personal data.
• The employer’s contact information.

Employers and developers need not disclose trade secrets or other legally protected information, but must notify the affected person if information is being withheld and why.

Enforcement

These requirements (and the developer obligations below) are enforceable solely by the Connecticut attorney general under the Connecticut Unfair Trade Practices Act (CUTPA); there is no private right of action. For violations through December 31, 2027, the attorney general may provide notice and a 60-day cure period before bringing an action if the violation is curable.

Developers’ obligations beginning October 1, 2027

SB 5 also covers developers: any person doing business in Connecticut that “develops, or intentionally and substantially modifies,” an AEDT advertised, marketed, configured, contracted for, sold or licensed to materially influence employment-related decisions. Developers must give employers all information needed to meet their disclosure and notice obligations, and the parties may agree by binding contract that the developer will assume specified obligations.

Colorado’s SB 26-189 imposes a similar information-sharing requirement. And because modifying a tool can trigger developer status, employers that heavily customize third-party tools should consider whether they could be treated as developers.

Use of AI is no defense to discrimination claims (effective October 1, 2026)

SB 5 amends the Connecticut Fair Employment Practices Act (CFEPA) to provide that use of an AEDT is not a defense to a discrimination claim, building on Attorney General William Tong’s February 2026 memo confirming that existing anti-discrimination laws apply to AI-driven decisions. Although anti-bias testing is not required, the Connecticut Commission on Human Rights and Opportunities (CHRO) or a court may consider evidence of such testing or “similar proactive efforts,” including their quality, efficacy, recency, scope, results and the employer’s response. As we reported in our September 2025 alert, California’s Civil Rights Council regulations, effective October 1, 2025, contain nearly identical language.

New AI disclosure for layoff notices (effective October 1, 2026)

Following New York’s lead, SB 5 requires employers that serve a federal Worker Adjustment and Retraining Notification (WARN) Act notice on the Connecticut Department of Labor to disclose, in a form the labor commissioner prescribes, whether the layoffs relate to the employer’s use of AI or “another technological change” (an undefined term). As we noted in a February 2025 alert, New York Gov. Kathy Hochul proposed a similar requirement. AI is defined more broadly than AEDT, as “any machine-based system that, for any explicit or implicit objective, infers from the inputs such system receives how to generate outputs … that can influence physical or virtual environments.” The requirement is disclosure only, but employers should carefully consider how they characterize the reasons for a reduction in force.

Other notable provisions

SB 5 also includes two other provisions of note for AI developers:

• Frontier developer whistleblower protections. “Frontier developers” (generally, those training foundation models using more than 10²⁶ integer or floating-point operations) may not adopt or enforce rules, policies or contracts permitting retaliation against employees for protected whistleblowing, including reports that the developer’s activities pose a specific and substantial danger to public health or safety due to a “catastrophic risk.” By January 1, 2027, “large frontier developers” (those with more than $500 million in annual gross revenues, including affiliates) must also establish an anonymous internal reporting process for covered employees. All frontier developers must notify covered employees of their rights. Violations carry civil penalties of up to $1,000 each, enforced by the attorney general.
• AI regulatory sandbox. Effective July 1, 2027, the commissioner of economic and community development must develop a plan for an AI regulatory sandbox allowing applicants to temporarily test innovative products or services under reduced regulatory requirements, with legislative recommendations due by January 1, 2028. The concept is similar to Texas’s sandbox under the Texas Responsible Artificial Intelligence Governance Act (TRAIGA), discussed in our September 2025 alert.

Next steps for employers and developers

Although the AEDT requirements begin October 1, 2027, the CFEPA amendment and WARN disclosure are already in effect. Employers and developers should consider the following steps now:

• Inventory AI tools. Employers should identify automated tools used in employment decisions and assess which may qualify as AEDTs, including tools that only inform a human decision-maker. Developers should identify products advertised, marketed or licensed for employment decisions that may qualify as AEDTs.
• Review vendor contracts. Employers should confirm vendors will supply the information needed for required disclosures and notices and consider allocating those obligations by contract, along with audit rights, indemnities and cooperation obligations. Developers should decide whether to assume any employer obligations, clearly document any such allocation and identify information they may withhold as trade secrets.
• Develop disclosure and notice templates. Employers should prepare plain-language AEDT interaction disclosures and tool-specific pre-decision notices that include each required element. Developers should prepare standardized documentation describing each tool’s purpose, the personal data it processes, the data’s sources and how it is assessed, so customers can readily complete their notices.
• Commission or review anti-bias testing. Employers should confirm their tools have been tested for bias, document the results and remediate any disparities identified. Because employers may rely on such testing to defend discrimination claims, developers should expect customers to request testing results and consider how to share them. Given California’s nearly identical standard, a single well-designed testing program may support compliance in both states.
• Update WARN notice procedures. Employers should promptly add the AI-related disclosure to their WARN Act checklists and forms and watch for the labor commissioner’s prescribed form.
• Train relevant internal teams. Employers’ human resources, talent acquisition and other relevant teams should understand that relying on AI does not insulate employment decisions from discrimination claims. Developers’ product, sales and marketing teams should understand that how a tool is marketed or licensed can determine whether SB 5 applies.
• Assess whistleblower obligations. Potential frontier developers should review policies and agreements for compliance with the anti-retaliation provisions and implement required employee notices; large frontier developers must also have an anonymous internal reporting process in place by January 1, 2027.

Bottom line

SB 5 adds Connecticut to a growing list of jurisdictions regulating AI in employment, with requirements that overlap with, but differ from, those in California, Colorado, Illinois and New York City. Multistate employers should build compliance programs that meet the most demanding requirements while accounting for state-specific differences. We will continue to monitor developments in this area.

Employers and developers with questions regarding the use of AI tools should contact their Cooley employment lawyer or one of the lawyers listed below.