Connecticut employers woke up on October 1 to a new legal reality: the algorithm is no longer an alibi.
Public Act 26-15, signed by Governor Ned Lamont in May, explicitly strips away the defense that a machine made the hiring decision.
The statute, formally titled the Connecticut Artificial Intelligence Responsibility and Transparency Act, amends the Connecticut Fair Employment Practices Act to declare that deploying automated employment-related decision technology is not a defense to a discrimination complaint. Proskauer Rose LLP noted the law applies broadly to any employer doing business in the state, employing Connecticut residents, or accepting applications from them.
“The AI did it” is not a defense. As of October 1, 2026, CFEPA says expressly that using an automated employment-related decision technology is not a defense to a discrimination complaint.
— Shipman & Goodwin LLP briefing to Connecticut Bar Association and CTCPA
There is a practical off-ramp, though it falls short of a safe harbor. The Commission on Human Rights and Opportunities or a court may weigh evidence of anti-bias testing — how rigorous, how recent, how broad, what it found, and how the employer responded. That documentation is not immunity, but it is a compelling reason to demand vendor testing results now and record what was done with them.
Notice rules arrive next October
Starting October 1, 2027, the transparency obligations switch on for tools deployed on or after that date. Employers must tell applicants and employees in plain language when they are interacting with covered technology, unless a reasonable person would find it obvious.
Before a tool serves as a substantial factor in an employment-related decision, a written notice must disclose the tool’s purpose, its trade name, the categories and sources of personal data it uses, and how that data is assessed.
The definitions carve out meaningful territory. Decisions about nonmaterial job changes, workplace health and safety, scheduling, planning, and productivity monitoring fall outside the notice mandate. Enforcement rests solely with the Attorney General; no private right of action exists, and a cure period runs through the end of 2027. Vendors can assume notice duties contractually, but the agreement must spell out exactly which duties they are taking on.
Shipman & Goodwin LLP said that the General Assembly may adjust the law in the upcoming legislative session.
Proskauer Rose LLP flagged a gray zone: the statute does not define what it means for an AI output to “meaningfully alter” the outcome of a decision, leaving the “substantial factor” standard open to regulatory interpretation. Employers withholding information as a trade secret must notify the affected individual of the withholding and the legal basis for it.
WARN notices now ask about AI
Already in effect since October 1, 2026, any employer filing a WARN notice with the Connecticut Department of Labor must disclose whether the layoffs are related to the use of AI “or another technological change.” It is a disclosure requirement only, but the language used to describe a reduction in force now carries strategic weight.
Colorado comparison and jurisdictional reach
Proskauer characterized the Connecticut law as narrower than Colorado’s recently enacted AI statute, yet its jurisdictional hooks are wide. Any employer with a Connecticut nexus — employees, applicants, or business operations — faces the same compliance calendar. The patchwork of state-level AI regulation is expanding rapidly, and Connecticut’s entry adds another layer for multistate employers to manage.
The first step is an inventory
For both legal counsel and finance leaders signing off on HR technology purchases, the immediate task is identical: map where AI already sits in hiring and personnel decisions. Most organizations find more than they expected. The law does not require perfection on day one, but it demands visibility — and a paper trail — before the next compliance deadlines arrive.