AI hiring laws in 2027: What Colorado, Connecticut, and California require

Starting January 1, 2027, Colorado employers must explain in plain language, within 30 days, how AI influenced an adverse hiring decision, and offer human review. Connecticut requires written AI notices from October 2027, and California limits AI-led discipline and firing from July 2027. Each law asks the same question: can you show how the decision was made?

For talent teams, that changes what "responsible AI" means in practice. Bias audits still matter. But the new rules also ask whether a person can explain a specific outcome, review it, and change it. This guide covers what each law requires, when it starts, and how to prepare.

This article summarizes public law text and law-firm analysis as of November 2026. It isn't legal advice. Talk to your counsel about how these laws apply to your organization.

Key takeaways

• Colorado's SB 26-189 applies to employment decisions made on or after January 1, 2027, with a plain-language explanation required within 30 days of an adverse outcome.
• Connecticut already treats AI as no defense to discrimination claims, and adds written notice rules on October 1, 2027.
• California's No Robo Bosses Act requires human corroboration for AI-assisted discipline and termination from July 1, 2027.
• The shared thread is explainability and meaningful human review, not only bias audits.
• Teams that inventory where AI influences hiring decisions now will be ready for each deadline.

Which states regulate AI in hiring in 2027?

Three states add new rules for AI in employment decisions in 2027, on top of rules already in force in New York City and Illinois.

Illinois also prohibits employers from using AI in ways that discriminate against protected classes and requires notice of covered AI use, according to BakerHostetler. In California, privacy regulations under the CCPA add rights for certain significant employment decisions from January 1, 2027.

What does Colorado's AI law require from employers?

Colorado's new law replaced SB 24-205, its 2024 AI act, before that law took effect. SB 26-189 was signed on May 14, 2026 and applies to consequential decisions, including employment, made on or after January 1, 2027.

For employers that use covered automated decision-making technology, the law sets out four main obligations, per the bill summary:

• Notice at the point of interaction: Tell people clearly when they are interacting with covered technology.
• An explanation after an adverse outcome: Within 30 days of a decision such as a rejection, give a plain-language description of the technology's role.
• Data access and correction: People can request their personal data and the correction of factually incorrect data used by the technology.
• Human review: People can request meaningful human review and reconsideration of an adverse outcome. Mayer Brown notes this applies to the extent commercially reasonable.

Developers of covered technology must give employers documentation of intended uses, known limitations, and instructions for appropriate use and human review. Both developers and employers must keep the records needed to show compliance for at least three years. The attorney general enforces the law, with a 60-day notice and opportunity to cure before acting.

Some details are still moving. At an October 26, 2026 hearing on the attorney general's proposed rules, business groups argued that the draft definition of "material influence" could reach decisions that a person ultimately makes, BakerHostetler reports. The attorney general has also agreed to defer enforcement while related litigation proceeds. Watch for final rules before January.

What changes in Connecticut?

Connecticut moved in two steps under Public Act 26-15, as Foley & Lardner explains.

Since October 1, 2026, relying on automated employment technology is not a defense to a discrimination complaint under Connecticut law. Evidence that an employer ran anti-bias testing or took similar steps may be considered.

From October 1, 2027, employers must tell employees and applicants, in plain language, when they are interacting with covered technology, unless that's obvious. When the technology may be a substantial factor in an employment decision, the employer must give written notice before the final decision that covers:

• That automated employment decision technology is being used
• Its purpose and the nature of the decision
• The technology's trade name
• The categories and sources of personal data it analyzes, and how that data is assessed
• The employer's contact information

The attorney general enforces the notice rules under the Connecticut Unfair Trade Practices Act. There is no private right of action, and for some violations through December 31, 2027, employers may get 60 days to cure.

Does California's No Robo Bosses Act cover hiring?

Not directly. SB 947, signed on September 30, 2026, focuses on discipline and termination. From July 1, 2027, a California employer cannot rely solely on an automated decision system to discipline or fire someone. If it relies primarily on one, a person must corroborate the output with other information, such as supervisor evaluations, personnel files, or work product. If the output can't be corroborated, the employer can't use it for that decision.

The employer must also give the affected employee a written, plain-language notice that a person reviewed the decision, with a contact for questions. Violations carry a civil penalty of $500 each. For hiring decisions, California's CCPA regulations are the closer fit, with rights to opt out of qualifying automated decision-making or receive information about how it was used.

What counts as an automated employment decision tool?

This is the question most teams will need to answer first, and each law draws the line in a different place:

• Colorado: technology that "materially influences" a consequential decision. Mayer Brown summarizes this as output that is a non-de minimis factor and affects the outcome, including by ranking, scoring, recommending, or classifying. Incidental, trivial, or clerical uses are excluded.
• Connecticut: technology whose output, such as a prediction, ranking, or score, is a "substantial factor" in an employment decision, meaning it meaningfully alters the outcome.
• California SB 947: obligations apply when an employer relies "solely" or "primarily" on an automated decision system for discipline or termination.
• NYC Local Law 144: tools that substantially assist or replace discretionary decision-making.

Scheduling tools, spreadsheets that need human analysis, and basic databases generally fall outside these definitions. Ranking, scoring, and screening tools are the ones to look at closely.

Why explainability is now a hiring requirement

NYC Local Law 144 asked employers to audit their tools and tell candidates about them. The 2027 laws go further. Colorado asks you to explain how a tool influenced one specific outcome. Connecticut asks you to name the tool and describe the data it uses. California asks a person to check the output against other evidence before acting on it.

Candidates already notice the gap. In an ICIMS and Kyle & Co. survey of 1,000 U.S. job seekers, only 32% of those who believed AI evaluated their applications said the employer clearly disclosed it, and 41% suspected AI was used without being told.

That has a practical consequence. If a tool ranks candidates on an opaque score, nobody on your team can write the explanation Colorado asks for, and the person "reviewing" the decision has nothing to review. Explanation and human review only work when the evidence behind a recommendation is visible.

How should talent teams prepare?

1. Inventory your tools. List every tool that ranks, scores, screens, or recommends candidates or employees, including features inside your ATS, assessments, and interview tools.
2. Map where AI influences decisions. For each tool, note which decisions it affects and in which states those candidates or employees are.
3. Name the reviewers. Decide who reviews AI-assisted outcomes, what evidence they see, and whether they can change the result.
4. Test an explanation. Pick a recent rejection and try to write a plain-language description of the tool's role. If you can't, start there.
5. Ask vendors for documentation. Request intended uses, known limitations, data sources, human review guidance, and bias testing results.
6. Update notices and records. Draft candidate and employee notices with counsel, and plan to keep records for at least three years.

For the broader framework, including ownership, vendor review, and monitoring, see our guide to AI governance in HR.

Where does a solution like Findem fit?

The common thread in these laws is evidence a person can examine. That's how Findem approaches AI in hiring. Findem's platform supports human judgment by making initial candidate recommendations and leaves final hiring actions to people. As our responsible AI page states, Findem is not intended to be used as an automated employment decision tool.

• Recommendations with the evidence attached: Findem's expert-labeled, evidence-backed signals are designed to provide explainability at every step, so a reviewer can see why someone was surfaced, not just a score.
• Human review by design: Model outputs are reviewed, approved, or overridden by people before they inform hiring decisions.
• Independent bias testing: In December 2025, Findem completed an independent bias audit of its applicant-matching practices with Warden AI. The audit found those practices met Warden AI's fairness thresholds across key groups, and the results are published on a live AI Assurance Dashboard.

No tool makes an employer compliant on its own. Your process, notices, and records still matter. But AI that shows its reasoning makes the explanation and review these laws describe much easier to deliver.

AI hiring laws FAQ

When does Colorado's AI law take effect for employers?

Colorado SB 26-189 applies to consequential decisions made on or after January 1, 2027, and employment decisions are consequential decisions under the law. It replaced SB 24-205, the AI law Colorado passed in 2024, before that law took effect.

Do employers have to tell candidates they use AI in hiring?

In more places, yes. Colorado requires notice when people interact with covered automated decision-making technology from January 1, 2027. Connecticut requires written notice before a final decision from October 1, 2027. NYC Local Law 144 and Illinois already require notice of covered AI use.

Does human review protect employers from AI discrimination claims?

Not on its own. Connecticut has said since October 1, 2026 that using automated employment technology is not a defense to a discrimination complaint, though evidence of anti-bias testing may be considered. Human review matters most when the reviewer can see the evidence and change the outcome.

What counts as an automated employment decision tool?

Each law draws the line differently. Colorado covers technology that materially influences a consequential decision, Connecticut covers technology that is a substantial factor in an employment decision, and NYC Local Law 144 covers tools that substantially assist or replace discretionary decision-making. Incidental or clerical uses are generally excluded.

Does California's No Robo Bosses Act apply to hiring?

SB 947 focuses on discipline and termination. From July 1, 2027, California employers cannot rely solely on an automated decision system for those decisions and must have a person corroborate the output when they rely on it primarily. Separate California privacy regulations cover other automated employment decisions.

Is NYC Local Law 144 still in effect?

Yes. New York City has enforced Local Law 144 since July 5, 2023. It requires an independent bias audit of covered tools, a public summary of the results, and notice to candidates and employees.