Connecticut’s Artificial Intelligence Responsibility and Transparency Act (SB 5) was signed on May 27, 2026. The first employer-facing piece took effect October 1, 2026: written WARN notices must now tell the Connecticut Department of Labor whether the layoffs relate to the employer’s use of AI or another technological change. The larger notice obligations for automated employment decision tools begin October 1, 2027. HR teams have roughly a year to inventory their tools and build a defensible process.
A law in phases: why October 1 is only the first step
When a state passes a broad AI law, the headline date rarely tells the whole story. Connecticut’s SB 5, formally the Connecticut Artificial Intelligence Responsibility and Transparency Act, is a good example. Governor Ned Lamont signed it on May 27, 2026, and parts of it became effective on October 1, 2026, which is why local outlets from the CT Mirror to Fox 61 ran explainers this week. But for employers, the heavier lift arrives a year later.
The law applies to employees and job applicants in Connecticut and creates obligations for employers that deploy what it calls automated employment-related decision technology. Employer-focused law firm alerts, including those from Proskauer and Kilpatrick Townsend, describe a two-step structure: a narrow, immediate obligation tied to layoffs, and a broader notice regime that starts October 1, 2027.
The conversation has not stopped at compliance, either. The Yankee Institute reported on the day the law took effect that lawmakers are already saying it does not regulate enough, which is a reminder that the framework could be amended before the 2027 obligations arrive. HR leaders should plan for the law as written and stay alert to change.
What took effect on October 1, 2026
The immediate employer requirement is specific. Beginning October 1, 2026, an employer that issues written WARN notices must disclose to the Connecticut Department of Labor whether the layoffs covered by the notice are “related to the employer’s use of artificial intelligence or another technological change.”
Three practical points follow from that sentence.
The trigger is a WARN notice. Employers that never reach the threshold for issuing written WARN notices are not facing this particular disclosure. But for organizations restructuring, especially those automating back-office or customer-service functions, this is a live question.
The test is “related to.” That phrase is broader than “caused by.” A reduction in force that follows the rollout of a new tool, even if multiple factors played a role, may require a thoughtful answer. The words “another technological change” widen the scope beyond generative AI to automation and software changes generally.
The disclosure is a statement to a state agency. It will sit alongside other representations the company makes about a layoff. Consistency matters: what you tell the Department of Labor should line up with internal business cases, board materials and communications to employees.
What arrives on October 1, 2027
Starting October 1, 2027, employers doing business in Connecticut, employers with employees in the state and employers that accept job applications from Connecticut residents must comply with notice requirements when they use an automated employment decision tool to make, or materially influence, employment decisions. Affected applicants and employees must receive written notice disclosing the use of the tool, along with other required disclosures set out in the statute.
The reach is wide. The statute is not limited to employers headquartered in Connecticut. Remote employers that accept applications from Connecticut residents are within the described scope. That is the same pattern seen in state privacy laws, and national employers should assume that Connecticut compliance will have to be built into their general hiring process rather than carved out for one state.
Law firm guidance repeatedly flags one analytical question as central: do AI outputs “meaningfully alter” employment-related decisions? If a recruiter uses a screening tool only as a convenience and makes every decision independently, the answer may differ from a workflow in which candidates below a score never reach a human. Documenting how outputs are used by human decision-makers is therefore not a formality; it is the evidence that supports your classification.
What the news coverage tells us about employer risk
A scan of the week’s coverage shows several recurring themes that HR teams can use as a risk lens.
- Vendor reliance is the common blind spot. A trade article in Insurance Business warned that many clients assume their technology vendor is handling compliance. Under laws like this one, a vendor’s assurances do not transfer the employer’s own notice obligations.
- The law sits in a larger compliance cluster. Connecticut’s AI provisions took effect alongside data-privacy changes, and commentary such as the JD Supra piece on Connecticut’s expanding privacy and AI compliance landscape treats them together. HR data, including applicant data, is often where privacy and AI obligations overlap.
- Employer cheat sheets are circulating. Fisher Phillips published a checklist of workplace laws taking effect in October 2026, a sign that firms expect employers to need practical, multi-jurisdiction trackers.
- Political attention remains high. The statement from lawmakers that the law may not regulate enough suggests that amendments or additional bills are plausible.
As for search behavior, Google Trends data for the United States on the analysis date was dominated by sports and entertainment, with no AI-employment query among the top trending searches. As with most compliance topics, attention is concentrated among professionals reading specialist outlets and law firm alerts, not among the general public. For employers, the implication is that nobody will remind you; the obligations arrive on schedule whether or not they are trending.
A twelve-month readiness plan for HR
Months 1 to 2: build the inventory
List every tool used in recruiting, screening, interviewing, performance management, scheduling, promotion and termination. Include tools embedded in applicant tracking systems and HR suites that you may not think of as “AI.” For each, note the vendor, what data it processes, what output it generates and who sees that output.
Months 3 to 4: classify by influence
For each tool, ask whether its output makes or materially influences an employment decision. Interview the people who actually use the tool. A policy that says “humans decide” means little if managers routinely accept recommendations without review. Record your reasoning and keep it.
Months 5 to 6: engage vendors
Request documentation of how each tool works, what it was trained or configured to do, what testing has been performed and what support the vendor will provide for employer notices. Update contracts to require cooperation, notification of material changes and reasonable audit rights.
Months 7 to 9: draft notices and processes
Prepare applicant and employee notice language, decide where in the process notices will be delivered, and design a channel for people to ask questions. Check the final statutory text and any agency guidance for required content before finalizing templates.
Months 10 to 12: train and test
Train recruiters and managers, run a dry-run of the notice workflow and confirm that records are retained in a way that would let you demonstrate compliance. Review the law again for amendments before the October 2027 date.
The WARN disclosure in practice: a short worked scenario
Imagine a Connecticut-based services company that deploys an automated claims-triage tool in spring and, in autumn, announces a reduction affecting its processing team. Leadership says the cuts reflect lower volume and a regional consolidation. Operations leaders, however, acknowledge in internal planning documents that the new tool cut manual handling time by a third. When the company prepares the written WARN notice after October 1, 2026, the statement to the Department of Labor must address whether the layoffs are related to the use of AI or another technological change.
A careful process would ask: what do our own documents say? Is the technology a contributing factor? Would a regulator reading our internal business case reach the same conclusion as our notice? The point is not that every layoff involving software must be labeled one way; the point is that the answer should be reasoned, documented and consistent. A disclosure that contradicts the company’s own planning materials invites scrutiny that a thoughtful one avoids.
Implications beyond Connecticut
Connecticut is not alone in moving on AI in the workplace, and employers operating across states face a patchwork. Rather than building separate programs for each state, many organizations are converging on a common baseline: a central inventory of AI tools, a standard impact assessment, a single notice framework adaptable by jurisdiction and a governance committee with HR, legal, IT and privacy representation. The Connecticut timeline, with a narrow obligation now and a broad one in 2027, gives a useful forcing function to build that baseline on a realistic schedule.
Smaller employers should not assume the law bypasses them. The notice obligations described in law firm summaries turn on using automated tools to make or materially influence decisions and on Connecticut connections, not on company size alone. Even a modest firm using an applicant tracking system with ranking features could be in scope.
Bottom line
October 1, 2026 delivered the first, narrow employer obligation under Connecticut’s AI law: telling the state whether a WARN-notice layoff is related to AI or another technological change. The main event, notices to applicants and employees about automated employment decision tools, arrives on October 1, 2027. The organizations that will handle it best are those that start now with an honest inventory, a clear view of how human decision-makers use tool outputs and vendor contracts that support compliance rather than assume it. Watch for legislative changes, read the statute and agency guidance directly, and treat this year as preparation time rather than waiting time.
Sources and further reading
- Proskauer, “Connecticut Enacts New AI Law: What Employers Need to Know.”
- Kilpatrick Townsend, “Connecticut Enacts New AI Legislation: What Businesses Need to Know About SB 5.”
- CT Mirror, Fox 61, Yankee Institute, JD Supra, Insurance Business and Fisher Phillips coverage (September 18 to October 2, 2026).
This article is for general information and is not legal advice. Consult qualified employment counsel about your obligations.
Discover more from Kurums | Business Intelligence
Subscribe to get the latest posts sent to your email.