Connecticut’s AI Hiring Law Is Live: Why the Compliance Bill Lands on You, Not Your Vendor

Connecticut’s AI hiring law took effect on October 1, 2026 — and if your company uses AI resume screening tools for small business hiring, or is thinking about it, one sentence in the state’s Artificial Intelligence Responsibility and Transparency Act matters more than everything else in it: the compliance burden falls on you, the employer, not the vendor that sold you the software. Your ATS contract and your HR platform’s marketing promises do not move your legal exposure anywhere. This guide explains what the law covers, which tools count, what’s in force now versus what arrives in 2027, and a 7-step checklist a small team can actually follow.

What Connecticut’s law covers (more than you’d think)

The law — formally the Artificial Intelligence Responsibility and Transparency Act (CAIA), signed as SB 5 — regulates “automated employment-related decision processes”: any technology that processes personal data and produces an output (a score, ranking, recommendation, or classification) that is a substantial factor in a hiring, promotion, discipline, or termination decision.

In practice, that covers:

• Resume screening software that ranks or filters applicants
• Candidate ranking algorithms inside your applicant tracking system (ATS)
• Recruiting chatbots that decide who moves forward
• Automated interview-analysis tools that score video interviews
• Performance-management and workforce-analytics systems shaping promotions or terminations

The critical detail: the law can capture a tool even when a human stays in the loop — if the AI’s output meaningfully shapes the decision, it counts. The Connecticut Business and Industry Association (CBIA) notes that employers are often surprised at how many AI tools are already embedded in their workflows, which is why an honest inventory is always step one.

The part most owners will miss: YOU are the “deployer”

Under the CAIA, the employer using the tool is the “deployer,” and the deployer carries the compliance obligation — not the developer that built the software. Per analysis from law firm Shipman and Goodwin, summarized by the CBIA: an employer that purchased or licensed a high-risk AI tool from a third-party provider does not transfer its legal exposure to that provider. The vendor contract does not move the obligation.

So when your screening vendor says “we handle compliance,” they may be handling their compliance — supplying you with documentation, which the law does require of developers. But the applicant notices, the risk-management policy, the bias testing, and the human-review process are yours. The one narrow exception: developers may contractually assume your disclosure duties. That makes what you ask for before you sign one of your most important compliance steps.

What’s live now vs. what arrives in 2027 (read this carefully)

Connecticut phased the law in, and the heaviest employer paperwork doesn’t land until October 1, 2027. But part of it is already live:

In effect since October 1, 2026:

• The core automated-employment-decision framework and the developer–deployer allocation of duties
• “The AI did it” is not a defense against a discrimination complaint under Connecticut’s anti-discrimination laws. If your tool produces a discriminatory outcome, courts may still consider your proactive anti-bias testing as a mitigating factor — the quality, recency, and scope of it
• A rule that employers filing federal WARN Act mass-layoff notices must disclose whether the layoffs relate to the employer’s use of AI
• Whistleblower protections for certain frontier-model developer employees

Arriving October 1, 2027:

• Written pre-decision notices: before any AI-influenced employment decision, you must disclose the tool’s use, its purpose and trade name, the categories and sources of personal data analyzed, how the data is assessed, and your contact information
• Disclosure when someone interacts directly with automated technology (unless obvious)

Enforcement: there’s no private right of action — only the Attorney General can enforce — and violations count as unfair or deceptive trade practices. For violations through December 31, 2027, the AG must give you a 60-day chance to cure before suing. That window closes at the end of 2027.

Translation for a small business: you have about a year before the paperwork obligations bite, but the anti-discrimination exposure is live now. Use the runway — don’t treat it as a free pass.

Which AI resume screening tools for small business hiring count — and what they cost

Any tool whose scoring, ranking, or filtering meaningfully shapes who you interview or hire is probably in scope. Here are the budget-tier tools small teams actually buy, with 2026 starting prices:

• Manatal — AI candidate scoring inside a full ATS; starts at $15/user/month. Often cited as the best SMB value because screening and ATS come in one product.
• Workable — semantic resume scanning, AI sourcing, structured interview kits; from about $299/month for AI-screening tiers.
• Classet — an “AI recruiter” that calls applicants within seconds and runs structured phone screens; from $249/month self-serve. Built for high-volume hourly hiring.

Confirm current pricing with each vendor before you buy.

One more warning: pasting resumes into ChatGPT or Claude for a ranked shortlist is the most common pattern at small firms — and the least governed. No audit trail, no documented rubric, no record of what criteria the model used. Under Connecticut’s definition it can still count as an automated employment decision process; it’s just one with zero documentation. Use a proper tool with logging, or at minimum keep a written record of your criteria and decisions.

Your 7-step compliance checklist (no HR department required)

Adapted from CBIA/Shipman and Goodwin guidance, sized for a team that hires a handful of people a year:

1. Run an AI inventory. List every AI tool touching employment decisions — ATS, HR platform, benefits tools, recruiting chatbots, and any general-purpose AI used for screening.
2. Classify each system. Does the tool’s output materially shape a decision? If yes, treat it as covered.
3. Write a one-page risk-management policy. Name who’s responsible, how you test for bias, how you monitor the tool, and what happens when something looks wrong.
4. Do a quick impact assessment per covered tool. Document its purpose, the data it processes, where discrimination risk could creep in, and your safeguards. Update when tools change.
5. Draft your notices now. Job postings, application pages, and handbooks should disclose where AI plays a role. The full pre-decision notice format is required from October 2027 — don’t start it next September.
6. Keep a real human in the loop. Build genuine human review into AI-influenced decisions and document it.
7. Train your people. Anyone who hires or manages the tools needs the basics.

And the golden rule from earlier: review your vendor contracts for what documentation and cooperation the provider is contractually obligated to supply. If a provider won’t put bias-audit results and technical documentation in writing, you’re carrying exposure the contract didn’t address. If you’re still early in adopting AI, our AI for small business implementation guide shows how to roll out AI tools with governance from day one. And when tools act more autonomously — screening, messaging, doing admin work on your behalf — hold them to the same standard you’d apply to any AI agent: approval gates, activity logs, no unsupervised spending or sending.

The national trend: Connecticut is the template, not the exception

Connecticut isn’t alone. New York City’s Local Law 144 has been the enforceable template: employers using automated employment decision tools must commission an independent annual bias audit, post a summary publicly, and give candidates at least ten business days’ notice before the tool is used — with penalties from $500 for a first violation to $1,500 per day for ongoing violations.

And scrutiny is accelerating. As BBC News reported on October 8, the New York City Council put executives from OpenAI, Anthropic, Google, and Meta under oath at a rare full-council hearing on October 5 to answer for AI safety and accountability — while Elon Musk’s SpaceXAI was the only major company subpoenaed that did not appear, prompting the Council Speaker to say she would seek a court order to enforce the subpoena. The Council is weighing ten proposed AI bills, including one letting candidates opt out of AI deepfakes of themselves. Proposals, not law — but the direction is unmistakable, on both sides of the Atlantic.

For the wider state picture, see our state AI laws small business guide. For the European side, see our EU AI Act requirements for small business guide.

The compliance playbook — inventory your tools, test for bias, disclose to applicants, keep a human in the loop — is converging everywhere. Build it once and you’re ahead in every state.

5 questions to ask your hiring-software vendor today

Get these in writing before you sign or renew:

1. Will you supply bias-audit results and technical documentation for our impact assessments and disclosures?
2. Will you contractually assume our disclosure duties, or cooperate fully if we keep them?
3. Can you provide an audit trail — data used, score produced, reasoning — for each candidate?
4. What human-review controls prevent decisions on the tool’s output alone?
5. How do you handle trade secrets — and will you state in writing what you withhold and why?

If a vendor can’t answer these, you’ll be writing its compliance story yourself.

Frequently asked questions

The bottom line

Connecticut’s AI hiring law is live, and its message is direct: if you use AI to shape hiring decisions, you own how that AI behaves — not whoever sold it to you. The fix is mostly process, not money: an honest inventory, a one-page policy, basic bias testing, real human review, and vendor contracts that put promises in writing. The full notice duties arrive in October 2027, so you have a real runway — but the anti-discrimination exposure is already in force, and regulators from Hartford to New York City are clearly not done writing rules. Start the inventory this week.

This article is general information for small business owners, not legal advice. Employment law is fact-specific — talk to a qualified employment attorney about your situation before making compliance decisions.

Sources: CBIA employer guidance; Insurance Business / Shipman and Goodwin analysis; JD Supra SB 5 employer alerts; SpotSaaS 2026 tool comparison; Recruiterflow 2026 screening tools; Classet candidate screening software; BBC News on the Oct 5, 2026 NYC Council AI hearing; hearing recap.