For many years, the discourse surrounding artificial intelligence in the workplace advanced more quickly than any legislation could. Algorithmic screening tools were discreetly adopted by employers. Many companies have replaced human recruiters with resume-ranking systems. Additionally, employees frequently were unaware that they had been marked for termination by an automated system before a manager even called. Connecticut recently declared the end of the opacity era.
Public Act No. 26-15, a comprehensive law that imposes some of the strictest transparency requirements on AI-assisted employment decisions anywhere in the United States, was signed into law by Governor Ned Lamont on June 2, 2026. Artificial intelligence use by employers is not prohibited by law. However, it does mandate that they reveal when they use it, what information it processes, and how the results affect hiring, promoting, disciplining, or firing decisions. That is a significant change.
The law covers what it refers to as Automated Employment-Related Decision Technology, or AEDT, which is basically any system that uses personal data to generate rankings, recommendations, or predictions that have a significant impact on hiring decisions. Spreadsheets and other generic tools are separated out. However, it most likely qualifies if an employer uses an AI platform to rank job candidates or evaluate the performance of current employees. Employment law attorneys are already advising clients to assume their tools fall inside rather than outside of the purposefully broad coverage.
One clause that is receiving a lot of attention is that employers are now required to inform candidates and employees in plain language that an AI system was used before making any decisions. They must give the tool’s name, describe the data it used, and include contact details in case someone wishes to inquire further. On paper, this might seem simple. Building those disclosure pipelines before the October 2027 deadline will actually require significant operational work for businesses that have implemented several overlapping AI systems across HR workflows.
The anti-discrimination section also contains something noteworthy. The law explicitly states that using an AI system is not a legal defense if the system results in discriminatory outcomes, amending Connecticut’s current civil rights statute. Although this has long been implied in federal employment law, putting it explicitly in a statute makes it more obvious to employers that they cannot simply blame the algorithm. The business is still accountable if the tool had an uneven effect on protected groups. However, when evaluating a discrimination claim, courts may take into account evidence of anti-bias testing, which essentially provides employers with a road map for creating a record that can be defended even before a lawsuit is filed.
Here, it’s difficult to ignore the meticulous construction. Formal third-party audits are not required by law. AI systems don’t need to be pre-approved before being deployed. In the absence of legislative action, a pilot program for independent verification organizations doesn’t even begin until July 2027 and ends in 2030. According to legal experts, it is a disclosure-focused strategy that takes cues from the EU and California without fully adhering to either model. Some worker advocates who wanted stricter restrictions might be irritated by that. However, it might also be the exact reason it passed.
Colorado, Illinois, New York City, and California have all taken steps to regulate employment AI, so Connecticut is not the first state to do so. Here, the scope is different. Discipline and termination decisions are now explicitly covered, going far beyond the hiring context that the majority of previous laws addressed. Additionally, employers must indicate whether mass layoffs are related to AI or other technological advancements in WARN Act notices starting in October 2026. This is a truly novel requirement at the state level.
It’s unclear if other states will follow. There’s a feeling that other legislatures and the tech firms whose AI products HR departments have been purchasing in large quantities will be keeping a close eye on Connecticut’s outcome. Individual employees have no private right of action under the law, which grants the Connecticut Attorney General exclusive enforcement authority. For the time being. If the political appetite increases in subsequent sessions, that balance may change.
The practical implication for state employees is that you now have a legal right to know if a machine significantly influenced a decision that impacted your job. Six months ago, this right was nonexistent. Enforcement and how honestly employers define what constitutes a “substantial factor” will determine whether it has any practical significance. Both of those are still genuinely unclear.
