Connecticut CART Act Takes Effect, Marking the First U.S. Mandatory Disclosure Requirement for AI Layoffs

On October 1, 2026, the core provisions of Connecticut's AI Responsibility and Transparency Act (CART) officially took effect. Employers subject to the federal WARN Act (those with 100 or more employees) that conduct mass layoffs must disclose to the state Department of Labor whether the layoffs were caused by the use of AI or technological change.

Factual Background

According to publicly available information, the AEDT framework that took effect on the same day under the CART Act establishes the enforcement structure and specifies that “the use of AI tools does not constitute a defense against employment discrimination.” This is the first state legislation in the U.S. to include AI-caused layoffs in mandatory reporting. The legal community has warned that companies are not yet prepared for compliance, marking a milestone moment in which AI employer liability moves from advocacy to hard constraints.

This effective date superimposes the original WARN Act layoff notification obligation with AI-specific disclosure requirements, creating a dual compliance threshold. Companies must add an additional AI or technology factor label to their existing mass layoff reports, or they may face direct review by the state Department of Labor. The simultaneous implementation of the AEDT framework provides a closed-loop path from information collection to potential penalties, making the principle that “the use of AI tools does not constitute a defense against employment discrimination” a hard constraint in actual practice, rather than an abstract idea.

Mechanism Breakdown

The CART Act combines the existing WARN Act framework with AI disclosure obligations. Employers must clearly indicate AI or technological change factors in layoff filings. The AEDT framework is rolled out in tandem, building a complete chain from reporting to enforcement. This design directly converts the discussion of “AI may replace jobs” into a compliance requirement that “AI actually replacing jobs must be substantiated.”

Specifically, the WARN Act originally required employers with 100 or more employees to notify employees and the state government 60 days before layoffs. Now CART adds an AI attribution label on top of this, forcing companies to keep traceable records along their internal decision-making chain. The AEDT framework, through standardized reporting forms and subsequent review procedures, turns this label into quantifiable and traceable enforcement data. Once the two are combined, any attempt to package the use of AI tools as a “routine technology upgrade” will face evidentiary pressure. Companies must complete a causal relationship statement at the reporting stage; otherwise, subsequent enforcement may directly cite the provision that “the use of AI tools does not constitute a defense against employment discrimination” to rebut them.

In essence, this mechanism transforms AI deployment decisions from internal management matters into externally auditable matters. The reporting process requires companies not only to explain the scale and timing of layoffs, but also to break down the specific use cases of AI tools in recruitment, performance evaluation, task allocation, and other areas, and to explain how closely they are connected to layoff decisions. The enforcement chain provided by the AEDT framework ensures that this information does not remain on paper; instead, it may trigger on-site inspections or data comparisons, thereby amplifying compliance costs.

Industry Impact

After the law takes effect, companies with 100 or more employees will face additional reporting processes, and the legal community cautions that inadequate compliance preparation may create enforcement risk. The use of AI tools in recruitment, performance, and other areas will also face stricter scrutiny, and companies need to reassess the transparency of automated decision-making.

For manufacturing and technology services, this impact is especially direct. Companies that previously relied on AI to optimize workforce allocation must now disclose technology substitution pathways in layoff filings, which in turn affects their communication strategies with suppliers and investors. The risk of inadequate compliance preparation repeatedly emphasized by the legal community means that some companies may be forced to delay the deployment of AI projects due to missing internal processes, in order to avoid triggering mandatory disclosure. The use of AI tools in recruitment and performance is also under scrutiny, and companies need to establish more granular logging systems to record the handoff points between algorithmic inputs and outputs and human decisions, in order to address potential transparency reviews.

Overall, the new reporting process will increase the administrative burden on business operations; especially for multi-state employers, Connecticut's single-state requirement may become a reference point for a national compliance template. The rise in transparency requirements for AI tools will also prompt companies to redesign automated decision-making interfaces to ensure their outputs can be understood by external audits, rather than only serving internal efficiency.

Strategic Assessment

[Analysis, not fact] Based on the existing legislative path, Connecticut's approach may become a precedent for other states to reference; rising compliance costs may prompt companies to adjust the pace of AI deployment, but the specific effects will still require further observation.

At the strategic level, this precedent effect may gradually spread through interstate legislative competition. If other states emulate the combined model of CART and AEDT, they will force national companies to establish uniform AI layoff disclosure standards, creating a trade-off between compliance costs and the speed of technology deployment. Companies may choose to slow high-risk AI applications in the early stage, instead prioritizing internal audit and documentation systems to reduce uncertainty in multi-state operations.

Meanwhile, rising compliance costs will lead companies to reassess how they calculate ROI for AI projects, incorporating disclosure obligations and potential enforcement risks into upfront assessments. The current lack of preparation warned about by the legal community suggests that in the short term some companies will adopt a wait-and-see strategy, waiting for the first enforcement cases to clarify the boundaries before deciding whether to accelerate automation. In the long run, such hard constraints may push AI vendors to develop more interpretable tools to help customers meet reporting requirements, but the specific effects will still require further observation.