What Illinois HB 3773 actually changed, and when

If your company has even one employee in Illinois and you touch any AI tool during hiring, promotion, or discipline, the law changed on January 1, 2026, whether you noticed or not. HB 3773 amends the Illinois Human Rights Act (775 ILCS 5) to explicitly cover AI-driven discrimination in employment decisions. It's not a new standalone AI statute sitting off to the side. It's a change to the state's core civil rights law, which matters more than it sounds like it should.

Governor Pritzker signed the bill on August 9, 2024, giving employers roughly seventeen months of runway before the effective date, as confirmed by Seyfarth Shaw's client alert. Most employers used that runway to do nothing, because the law felt theoretical until the Illinois Department of Human Rights started drafting the rules that would operationalize it. That's where things got messy, and we'll get to that.

Because HB 3773 lives inside the Human Rights Act, a violation isn't a regulatory fine capped by statute the way some state AI laws are. It's a civil rights violation, which means the remedies on the table include back pay, attorney's fees, and uncapped compensatory damages through the Illinois Human Rights Commission. That's a meaningfully different risk profile than a notice-and-cure penalty structure, and it's the detail most employers researching this law skip past.

Who has to comply

If you employ one or more people physically located in Illinois, you're covered, even if your headquarters, your HR team, and your hiring managers sit in another state entirely. The Association of Corporate Counsel's FAQ guide confirms this scope explicitly: remote workers count, and out-of-state employers with Illinois-based staff are covered the same as a company headquartered in Chicago.

This catches a lot of mid-market companies off guard. A 200-person software company based in Austin with four remote engineers in Illinois is a covered employer. A staffing firm placing contractors with an Illinois client is covered. If you've expanded hiring geographically over the last few years, which most growing companies have, there's a decent chance you're in scope without having thought about it.

The decisions covered are broad too: recruitment, hiring, promotion, renewal of employment, selection for training or apprenticeship programs, discharge, discipline, tenure, and other terms and privileges of employment. That's effectively the full employee lifecycle. If your AI tool touches resume screening, interview scoring, performance review synthesis, or disciplinary recommendations, it's in scope.

What actually counts as a violation

The standard is impact, not intent. If an AI or generative AI system you use has a discriminatory effect on a protected class, "the algorithm did it, not us" is not a defense, and neither is "we didn't mean for that to happen." This mirrors how disparate impact theory has always worked under civil rights law; HB 3773 just makes explicit that algorithmic tools are not exempt from that theory.

The bill also includes one specific, concrete prohibition that's worth memorizing: employers cannot use zip code as a proxy for protected characteristics in AI-driven employment models, per the bill text on LegiScan. This sounds narrow, but zip code is one of the most common proxy variables that creeps into scoring models, whether through commute-distance logic, regional salary benchmarking, or location-based risk scoring baked into a vendor's black-box algorithm. If your screening tool was built by someone else and you've never asked what features it weighs, you likely don't know if this applies to you.

What's notably absent from the statute itself is a bias-audit mandate like NYC's. Illinois didn't write in a requirement for an independent annual audit. Instead, it leans on the existing enforcement mechanism of the Human Rights Act: file a complaint, IDHR investigates, and the employer has to show the tool wasn't causing discriminatory impact. That's a reactive enforcement model, which in practice means your first real test of compliance might be a charge, not an audit.

The notice requirement, in practical terms

Employers must notify applicants and employees when AI is used to influence or facilitate a covered employment decision. The statute's notice language is intentionally broad, which is both a gift and a headache: broad enough that most disclosure language will technically satisfy it, vague enough that nobody can tell you the exact required wording with confidence right now.

Hinshaw & Culbertson's analysis notes that IDHR was tasked with adopting rules to define the specifics, including what the notice needs to say and how it needs to be delivered. IDHR released draft Subpart J rules in December 2025 that would have clarified this, then temporarily withdrew the rulemaking hearings in June 2026 to coordinate with other state agencies, according to JD Supra's coverage of the postponement.

That withdrawal is the single most important nuance in this entire topic, and it's the thing most competing explainers bury or skip. The statute is live. The obligation to give notice is live. The detailed rules describing exactly how to satisfy that obligation are not finalized. That's not a loophole, it's a gap, and gaps in employment law enforcement tend to resolve in favor of whoever documented good-faith effort before the rules landed. Practically, that means: write a clear, specific notice now (what tool, what decision it touches, how it's used), log when it was given and to whom, and update it the moment IDHR finalizes Subpart J. Waiting for perfect clarity before acting is the wrong instinct here.

Two Illinois AI laws, not one

People conflate HB 3773 with the Artificial Intelligence Video Interview Act constantly, and it's worth being precise because they impose different obligations. The Video Interview Act (820 ILCS 42) has been in effect since 2020, well before HB 3773 existed. It applies narrowly: when you use AI to analyze video interviews of applicants for Illinois-based positions, you must notify the applicant before the interview, explain how the AI works and what characteristics it evaluates, and obtain consent before using the technology.

HB 3773 is broader and newer. It covers any AI use across the employment lifecycle, not just video interview analysis, and it's grounded in anti-discrimination law rather than a standalone notice-and-consent regime. If you only ever addressed the Video Interview Act (say, by adding a consent checkbox to your interview platform years ago), that compliance work does not carry over to HB 3773. You need both, and they require different documentation.

Where Illinois sits in the state-by-state patchwork

There is no federal AI employment law. The EEOC has issued guidance on AI in employee selection procedures, but it doesn't preempt or replace state law, which is exactly why a growing company now has to track obligations state by state. Illinois joins Colorado as one of only two states explicitly regulating algorithmic discrimination in private-sector AI use as of their respective effective dates, per Employment Law Worldview's analysis. NYC and California take different regulatory approaches entirely.

Here's how the four most relevant frameworks compare for an employer operating in multiple jurisdictions:

If you operate in more than one of these states, don't build four separate compliance programs. Build one standard that satisfies the strictest common denominator (usually that means audit-ready documentation plus explicit notice), and layer in the jurisdiction-specific wording where required. Our breakdowns of NYC Local Law 144, Colorado's AI Act, and California's ADMT regulations go deeper on each if you're building that single standard.

Why a rented, black-box AI hiring tool makes this harder to prove

Here's the uncomfortable part most vendors won't say out loud: if you can't explain how your screening tool scores a candidate, you can't produce the audit trail this law effectively demands, even though the statute doesn't use the word "audit." When IDHR or a plaintiff's attorney asks why an applicant was filtered out, "our vendor's model did it" is not an answer that holds up, and most off-the-shelf ATS screening add-ons were never built with that question in mind.

This is the same diagnosis problem we see across every regulated industry we work in. A tool can work fine in production and still be a compliance liability, because "works" and "explainable" are different properties. We've seen this pattern in healthcare operations, in financial services underwriting, and now in HR tech: the vendor optimized for accuracy and speed, not for the ability to reconstruct, six months later, exactly why a specific decision was made for a specific person. That reconstruction capability is what Illinois (and Colorado, and NYC) are actually asking for, whether or not the statutory language spells it out plainly.

Owning the system, or at least having real visibility into its scoring logic and training data, is the only way to answer these questions with confidence instead of a vendor's support ticket queue. That's a separate conversation from whether the tool works well. A tool can be both effective and undocumented, and undocumented is what gets companies sued.

What to ask before you keep using (or buy) an AI hiring tool

Run this checklist against whatever is currently touching your recruiting, screening, or promotion decisions:

• Can the vendor tell you, in plain language, which candidate or employee attributes the model weighs, and can you confirm zip code (or a close proxy like commute distance or regional cost-of-living index) isn't one of them?

• Does the tool log enough detail to reconstruct, for any individual decision, what inputs led to what output, six months after the fact?

• Has the tool been tested for disparate impact across protected classes, and how often, and by whom (internal team, vendor, or independent third party)?

• Does your current candidate-facing notice actually name the tool, describe what decision it influences, and get delivered before that decision is made, not buried in an onboarding packet after the fact?

• Does your vendor contract give you the right to audit the model, or are you contractually blind to how it works, which is common with SaaS ATS add-ons?

• If IDHR finalizes Subpart J with stricter notice or documentation requirements, can your current setup adapt, or would it require rebuilding the integration from scratch?

If more than two of these get an honest "no" or "we'd have to ask the vendor," you're carrying more exposure than the subscription price suggested. This is also the moment to pull in whoever owns your broader AI governance posture, not just HR, because the same explainability gap tends to show up anywhere a black-box model touches a regulated decision. Our piece on what AI agents actually do inside HR operations covers how these systems get built when the goal is both functional and auditable from day one, and our vendor risk assessment checklist walks through the exact contract terms to push for before you sign anything new.

If you're working through whether to patch your current vendor relationship, replace it, or build something your team actually owns and can explain, that's precisely the kind of diagnostic conversation we run before recommending anything, and it applies well beyond hiring to any AI system touching a regulated decision. Our enterprise AI services page covers how we structure that assessment, and we're happy to compare notes if you're mid-decision.

Frequently asked questions

Does Illinois HB 3773 apply to companies outside Illinois with remote employees there?

Yes. The law covers any employer with one or more employees physically located in Illinois, regardless of where the company is headquartered. A business based in another state with even a handful of remote Illinois employees is a covered employer under the amended Human Rights Act, per the ACC's FAQ guide on the law's scope.

Is Illinois HB 3773 actually enforceable now that the implementing rules were postponed?

Yes. The statute took effect January 1, 2026, and remains fully enforceable. IDHR withdrew its draft Subpart J rulemaking hearings in June 2026 to coordinate with other agencies, which delays the detailed compliance procedure, not the underlying legal obligation. Employers should comply with the statute's plain text now rather than wait for finalized rules.

What's the difference between Illinois HB 3773 and the Illinois AI Video Interview Act?

The Video Interview Act (820 ILCS 42), in effect since 2020, narrowly covers AI analysis of video interviews, requiring notice, an explanation of the technology, and consent. HB 3773 is broader, covering AI use across the full employment lifecycle (hiring, promotion, discipline, discharge) under anti-discrimination law. They're separate statutes with separate requirements, and compliance with one doesn't satisfy the other.

What notice does an employer have to give before using AI in hiring in Illinois?

Employers must notify applicants and employees when AI is used to influence or facilitate a covered employment decision. The statute doesn't yet specify exact wording or delivery mechanics, since IDHR's implementing rules are still in progress. The safest approach is a clear, specific written notice naming the tool and the decision it affects, delivered before that decision is made.

How is Illinois HB 3773 different from NYC Local Law 144 or Colorado's AI Act?

NYC Local Law 144 requires an independent annual bias audit with published results before using an automated employment decision tool. Colorado's AI Act requires a broader risk management program and impact assessments for high-risk AI systems. Illinois HB 3773 takes an impact-based civil rights approach with a notice requirement and a specific zip-code-proxy ban, enforced through the Human Rights Commission rather than a dedicated AI regulator.