If you use any AI in hiring, promotion, or discipline decisions that touch Illinois workers, HB 3773 already applies to you. It took effect January 1, 2026, and it reaches any company with even one employee in Illinois. When the Illinois Department of Human Rights (IDHR) withdrew its proposed implementing rules on June 2, 2026, that did not pause the law. It just means you comply from the statute itself while the detailed rules are rewritten.
What Illinois HB 3773 Requires
HB 3773 (Public Act 103-0804) amends the Illinois Human Rights Act (IHRA) to make employer use of AI in employment decisions an explicit civil rights issue. Violate it and you face the same enforcement and remedies as any other IHRA discrimination complaint. The IHRA covers employers with one or more employees, across the full employment lifecycle from recruitment and hiring through promotion, discipline, and discharge.
The statute creates three independently actionable violations: using AI that has the effect of discrimination against a protected class, using ZIP codes as a proxy for a protected class, and failing to give required notice when AI is used. Each stands on its own, so an applicant could file a complaint simply because you never disclosed that AI screened them, even if the AI produced no discriminatory outcome. For how Illinois rules intersect with screening, see the Illinois employment screening compliance guide.
How Broadly HB 3773 Defines AI
Broadly enough that most HR technology likely qualifies. If a tool uses machine learning, natural language processing, predictive analytics, or generative AI anywhere in a covered decision, assume it is in scope. Any "use" of AI triggers the law, so it does not matter that AI is one factor among many or that a human reviews the output first. Resume-screening software, AI chatbots in hiring workflows, AI-assisted background check adjudication, performance-scoring systems, and generative AI that summarizes candidates all typically qualify. If you are unsure where AI ends and a consumer report begins, this piece on when an AI hiring score becomes a consumer report is worth your time.
The Notice Requirement
Failing to give AI notice is its own civil rights violation, separate from whether the AI discriminates. You must notify both applicants and current employees whenever AI influences or facilitates a covered decision, and that trigger is deliberately low. A workable notice states plainly that AI is being used, identifies the stages it applies to, describes in general terms what the AI evaluates, and says whether a human also reviews the decision, in language a non-specialist can follow. If you run AI video interviews, the Illinois Artificial Intelligence Video Interview Act already requires advance notice, an explanation, and applicant consent, and HB 3773 stacks on top. Build the notice into portals, e-signature packets, handbooks, and promotion and discipline procedures, and log when each was shown and acknowledged. For how bias and employer obligations intersect, see the post on AI hiring bias and compliance.
The Rulemaking Delay: What It Means Now
IDHR published proposed rules on May 15, 2026, then withdrew them on June 2, 2026, citing a need to keep working with other state agencies. That leaves operational details unsettled, including the exact format for AI notices, specific delivery methods, and record-keeping requirements. What the withdrawal does not do is suspend the law: all three obligations became effective January 1, 2026, and remain enforceable now. Build your framework on the statutory text, err toward more thorough notice, and set a monthly reminder to check the IDHR rulemaking calendar so revised rules land as refinements, not rebuilds.
The HB 3773 Compliance Checklist
Work through this by function so you can assign each item to the right owner.
Governance and inventory
- Map every AI tool used across recruitment, application intake, background screening, interviewing, promotion, discipline, and termination, including tools that "just help" rank, schedule, or summarize.
- Classify each tool under the broad AI definition. When in doubt, treat it as covered.
- Assign clear compliance ownership across HR, legal, and IT, and bring your background screening provider into that framework.
Non-discrimination controls
- Remove ZIP code, neighborhood, and similar geographic variables from any model influencing covered decisions, unless a variable is strictly job-related and you can document why.
- Review other inputs that may proxy for a protected class, such as certain schools, credit indicators, or neighborhood-linked criminal data, and document your rationale for keeping any that raise questions.
- Require fairness documentation from every AI and screening vendor: inputs, outputs, testing, and a written commitment to comply with the IHRA. The Mobley v. Workday litigation shows why vendor tools can still land on your desk.
- Keep trained HR staff reviewing high-risk, AI-flagged decisions, especially "do not hire" recommendations that follow a background check.
Notice
- Draft plain-language AI notices and pair them with Video Interview Act disclosures where they apply.
- Integrate notices into every applicant and employee touchpoint, from portals and e-signature packets to handbooks and review processes.
- Configure your systems to record when notices are shown and acknowledged, and retain those records.
Policy, training, and monitoring
- Update EEO and anti-discrimination policies to name AI, and add an HR AI-use policy aligned with HB 3773.
- Train recruiters and managers on when AI is used, how notice must be delivered, and what counts as prohibited use.
- Run periodic internal reviews of AI-driven outcomes across demographic groups, and update your IHRA complaint response so it can pause a problematic tool and reevaluate affected decisions.
What This Means for Your Background Screening Program
Much of a typical screening workflow already involves AI: automated risk scoring, adjudication engines, AI-assisted interpretation, and generative AI report summaries all fall within scope. Start with your vendor contracts, adding representations that the AI complies with the IHRA and embeds no prohibited proxies. Coordinate AI notice with your existing FCRA and state disclosures, adding Illinois-specific language that automated tools may assist in interpreting results. Above all, keep a qualified person in the loop on adverse decisions: if AI recommends "do not hire" off a background check, a human should review before any adverse action, a safeguard under both HB 3773 and FCRA. KRESS employment screening services are built to show employers where AI touches their workflow and to build compliance in from the start. For the wider picture, see the five AI hiring laws that could catch your company off guard.
Frequently Asked Questions
Is HB 3773 enforceable now even though the IDHR rules were withdrawn?
Yes. The statute took effect January 1, 2026. The June 2, 2026 withdrawal only postpones detailed operational guidance; the core prohibitions and the notice duty are active, enforceable law. It also means the burden of showing non-discrimination sits with you once a complaint is filed, even though HB 3773 does not mandate a formal bias audit like New York City's Local Law 144.
Do I need to give AI notice to current employees, or only job applicants?
Both. Notice is required whenever AI is used in any covered decision, including promotion, discipline, performance evaluation, training selection, and discharge.
Does HB 3773 apply if AI only assists a human decision-maker?
Yes. The law covers AI that influences or facilitates decisions, not only fully automated ones. Using AI as one input, or as a recommendation a human reviews, still triggers both the notice duty and the non-discrimination prohibition.
Can my screening provider's AI tools create liability for my company?
Yes. Employers are responsible under the IHRA for AI used in their employment decisions, whoever supplies it. If your provider uses AI-assisted adjudication, risk scoring, or report summaries, confirm what the AI does, that it uses no prohibited proxies, and that your notices disclose it.
What is the ZIP-code proxy ban?
HB 3773 expressly prohibits using ZIP codes as a proxy for a protected class in covered decisions. ZIP codes correlate strongly with race and national origin, so audit any model that embeds ZIP code, neighborhood, or similar location data.
The law is live, the rules are still forming, and you are the one who has to make it work inside your organization. You do not have to figure it out alone. KRESS can pinpoint where AI touches your hiring and screening, flag the gaps, and build compliance in before a complaint forces the issue. Reach out for a screening program review and start from what you are already doing right.