Connecticut AI Employment Law Takes Effect October 2027
New statute requires employers using automated decision tools to provide detailed notices to workers and job applicants.
Connecticut has joined the growing number of states regulating artificial intelligence in employment decisions. Governor Ned Lamont signed Senate Bill 5, the Connecticut Artificial Intelligence Responsibility and Transparency Act, into law on May 27, 2026, according to Law and the Workplace.
The legislation targets what it calls "automated employment-related decision technology" or AEDT—broadly defined as any technology processing personal data to generate predictions, recommendations, classifications, rankings, or scores that substantially factor into employment decisions. The law excludes basic productivity tools like word processors and spreadsheets, as well as systems used only incidentally in employment decisions.
Why it matters
Connecticut's law adds another layer to the fragmented state-level AI regulatory landscape, creating compliance challenges for multi-state employers. Unlike some state laws that focus solely on algorithmic bias, this statute combines transparency mandates with anti-discrimination provisions, requiring employers to document proactive testing efforts that could influence future litigation outcomes.
Notice requirements begin October 2027
Starting October 1, 2027, employers deploying AEDT that interacts with employees or applicants must disclose that interaction in plain language—unless a reasonable person would find it obvious they're engaging with automated technology.
When AEDT generates outputs that serve as a "substantial factor" in employment decisions, employers must provide written notice before the decision occurs. The statute defines substantial factor as anything that "meaningfully alters the outcome" of an employment decision, though it leaves that phrase undefined—likely creating interpretive questions for regulators and courts.
Required disclosures include the AEDT's trade name, its purpose, what categories of personal data it analyzes, data sources, and employer contact information. Trade secrets remain protected, but employers must notify individuals when withholding information on that basis.
The Connecticut Attorney General enforces these provisions as unfair trade practices. Violations occurring through December 31, 2027 trigger a notice-and-cure process with a 60-day window. The law explicitly prohibits private lawsuits.
Anti-discrimination amendments
Effective October 1, 2026, amendments to Connecticut's Fair Practices Act clarify that using AEDT provides no defense against discrimination claims. Employers cannot avoid liability by claiming they simply followed AI recommendations.
However, the Connecticut Commission on Human Rights and Opportunities and courts may consider evidence of anti-bias testing and similar proactive measures when evaluating claims. This creates an incentive for employers to conduct and document testing before deployment.
WARN Act disclosure
Also beginning October 2026, employers issuing Worker Adjustment and Retraining Notification Act notices must tell the Connecticut Department of Labor whether layoffs relate to artificial intelligence use or other technological changes.
Implementation considerations
Employers should inventory their AI tools to determine which ones "meaningfully alter" employment outcomes versus merely assisting with administrative tasks. The distinction will likely be fact-specific, depending on actual use rather than vendor marketing claims.
Because the law assigns different responsibilities to AEDT developers and deployers, employers should review vendor contracts to clarify which party handles notice obligations. Developers must provide information employers need for compliance, and contracts can explicitly shift certain duties to developers.
Given that anti-bias testing evidence may influence discrimination cases, employers should consider conducting such testing—or verifying that vendors have done so—before deploying covered systems.
These details were first reported by Law and the Workplace, authored by Guy Brenner and Delia Karamouzis of Proskauer Rose LLP.
This is an original analysis by the Omega editorial team. Source reporting: AI Watch.
Want systems like this working for your business?
Book a Call