“Connecticut Enacts Comprehensive Artificial Intelligence Law: What Employers and Businesses Need to Know,” by Dove A.E. Burns, Esq., and Stacey Pitcher, Esq., 6-30-2026

Connecticut has joined a growing number of states regulating artificial intelligence by enacting Connecticut Public Act 26-15, commonly referred to as the Connecticut Artificial Intelligence Responsibility and Transparency Act (CART Act). The new law establishes a broad framework governing the development, deployment, and use of artificial intelligence and imposes new compliance obligations on covered businesses.

The CART Act addresses a wide range of AI-related activities, including automated employment-related decision technologies, artificial intelligence companions, frontier AI models, generative AI content provenance requirements, and protections for minors interacting with certain AI systems.

As organizations increasingly integrate AI into workplace operations, customer-facing functions, and other business activities, the CART Act reflects the growing trend toward state-level regulation of artificial intelligence.

The Act contains staggered effective dates, with certain provisions taking effect in July 2026 and additional substantive requirements becoming effective beginning in October 2026 and throughout 2027, making it important for businesses to review the law’s requirements carefully to determine which obligations apply and when compliance is required.

Employment-Related AI Systems

One of the most significant aspects of the CART Act for employers is its regulation of automated decision systems used in employment-related decisions.

The law establishes notice and transparency requirements for certain automated employment-related decision technologies used in connection with hiring, promotion, discipline, discharge, renewal of employment, selection for training or apprenticeship opportunities, and other decisions affecting the terms, privileges, or conditions of employment. More precisely, the statute focuses on “automated employment-related decision technology” and requires disclosures to employees and applicants when such technology processes personal data and uses computation to generate an output—such as a prediction, recommendation, classification, ranking, or score—that is a substantial factor used to make, or that materially influences, an employment-related decision.   Beginning October 1, 2027, employers using this technology must provide written pre-decision notice stating: (a) that the business used the technology; (b) the purpose of the technology an the nature of the employment-related decision; (c) the trade name of the technology; and (d) the categories and sources of personalized data that will be analyzed and how it will be assessed.

The Act reflects increasing legislative scrutiny of workplace AI technologies and follows a broader national trend toward regulating the use of automated decision-making systems in employment. The statute also makes clear that use of such technology is not a defense to discrimination complaints.

Consumer Transparency Requirements

The CART Act also establishes transparency obligations for businesses that deploy AI systems when interacting with consumers.

Among other requirements, the law regulates certain “artificial intelligence companions” and requires clear disclosures when users could reasonably believe they are interacting with a human being rather than an AI system. These provisions are intended to promote informed consumer engagement and reduce the risk of deceptive or misleading interactions.

Frontier Models and Generative AI

The legislation also includes provisions directed at developers of advanced AI models, commonly referred to as frontier models. These provisions are intended to address catastrophic-risk concerns associated with highly capable AI systems and establish compliance obligations for certain frontier developers.

The statute is more specific here: it creates duties for “frontier developers” and “large frontier developers,” including internal reporting, notice obligations, and protections related to catastrophic-risk concerns. Those provisions are not just general “responsible development” language; they are compliance rules tied to defined risk thresholds and defined developer categories.

The Act also establishes provenance-data requirements for certain covered providers of publicly accessible generative AI systems. Generative AI is defined in the statute to mean “any form of artificial intelligence, including, but not limited to, a foundation model, that is able to produce synthetic digital content. Subject to specified exceptions and technical limitations, covered providers must take commercially and technically reasonable steps to include provenance data in certain AI-generated or materially altered audio, image, and video content.

As generative AI technologies continue to evolve rapidly, these requirements reflect growing concerns among policymakers regarding transparency, authenticity, and accountability in AI-generated content.

Protections for Minors Online

The CART Act includes special requirements applicable to operators of certain artificial intelligence companions when the operator knows or has reason to believe a user is under eighteen years of age. Among other restrictions, covered operators must implement safeguards designed to prevent AI companions from encouraging self-harm, violence, disordered eating, substance abuse, inappropriate emotional dependency, or romantic, erotic, or sexually explicit interactions with minors. The law also requires certain parental-control and account-management tools for minor users.

Authors: Dove A.E. Burns, Managing Partner of KD’s New Haven office and Regional Managing Partner of the firm’s California offices and Stacey Pitcher, New Haven Partner.