Connecticut Enacts New AI Transparency Requirements For Employers

Executive Summary: Effective October 1, 2026, Connecticut joins other jurisdictions, such as New York City, Illinois, and Colorado, in regulating Artificial Intelligence (AI) use in employment-related decisions. The new requirements are part of the larger Online Safety Act (OSA), Public Act 26-15, which was signed into law by Governor Ned Lamont on May 27, 2026. Among other things, the OSA requires that employers using "automated employment-related decision technology" in employment-related decisions disclose that information to applicants and employees starting on October 1, 2027.

Key Provisions

The employment-related provisions of the OSA regulate the use of "automated employment-related decision technology," broadly defined as “any technology that processes personal data and uses computation to generate any output, including, but not limited to, any prediction, recommendation, classification, ranking, score or other information, that is a substantial factor used to make or materially influence an employment-related decision.” It does not include common software used in business, such as word processors, spreadsheets, databases, and cybersecurity measures. An “employment-related decision” includes hiring, firing, promotion, discipline, selection for training or apprenticeship opportunities, tenure, employment renewal, and decisions related to other privileges, terms, and conditions of employment. PA26-15 specifically excludes “nonmaterial” workplace changes and decisions related to workplace health and safety, scheduling and planning, and productivity monitoring.

Beginning October 1, 2027, employers using “automated employment-related decision technology” must provide notice to applicants and employees under 2 circumstances: (1) where they will directly interact with the “automated employment-related decision technology; or (2) before an employment-related decision is made. In the first case, the notice must state in plain language that the person is interacting with automated technology. In the latter, notice must be in writing, and state:

• That “automated employment-related decision technology” is being used;
• The purposes of the “automated employment-related decision technology” and the nature of the employment decision at issue;
• The tradename of the “automated employment-related decision technology”;
• The category(ies) of personal data that the “automated employment-related decision technology” will use and how it will be assessed in reaching an employment-related decision;
• The source(s) of the personal data used; and
• The employer’s contact information.

The law further requires that any developers who market, advertise, configure, contract, sell, or license “automated employment-related decision technology” specifically for use in influencing employment-related decisions provide employers with all information necessary for the employer to comply with its disclosure obligations. Although trade secrets and other information otherwise protected by law are specifically exempted, the developer must inform the employer what information is being withheld and the basis for doing so.

Although no private cause of action for a violation is built into the law, the OSA amends the Connecticut Fair Employment Practices Act to preclude an employer’s reliance on AI as a defense to complaints alleging a discriminatory employment action, reflecting concern about algorithmic equations perpetuating discrimination against certain groups. It further authorizes the Connecticut Attorney General to bring claims under the Connecticut Unfair Trade Practices Act (CUTPA), Conn. Gen. Stat. § 42-110b, against an employer. However, for violations occurring between October 1, 2027 and December 31, 2027, the Attorney General may first issue a notice of violation and provide a 60-day cure period before initiating an enforcement action. In an enforcement action brought by the Attorney General for a violation of CUTPA, relief is generally equitable in nature, although civil penalties up to $5,000 per violation, or $25,000 per violation of an injunction, can be imposed.

The Bottom Line

Employers who use or plan to use AI to interact with applicants or employees or to make employment-related decisions should begin assessing whether their programs meet the definition of “automated employment-related decision technology.” Employers may need to work with vendors to ensure they have all the information needed to make proper disclosures. Importantly, employers should put in place policies and procedures to ensure that the required notice is given to applicants and employees who will be interacting with AI, and that written notice is provided to applicants and employees before employment-related decisions are made. Getting ahead of this now will make compliance effective October 1, 2027 easier.

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.