Illinois AI Employment Law: An Operating Playbook for HR Leaders

Why this is on your desk right now

HB 3773 has been in force since January 1, 2026. The Illinois Department of Human Rights is in the late stages of finalizing Subpart J — the rulemaking that will specify the timing, form, and recordkeeping requirements for AI-use notices in employment. The combination produces a window in which the law is operative against you today, the notice rules are not yet finalized, and the standard against which you will be judged is strict liability.

The audience for this playbook is HR leaders — CHROs, VPs of HR, heads of Talent Acquisition, HR Compliance officers, HRIS leads. The operational responsibility for HB 3773 lives with you. The legal framework and the board-level oversight question are covered in the HB 3773 Compliance Guide for Employers and the AI Risk and D&O Liability reference; this playbook is the operational layer.

Five things make HB 3773 different from the AI-employment laws HR leaders may be familiar with elsewhere:

• Strict liability. Intent is not a defense. An AI tool that produces a discriminatory outcome against a protected class is a violation even if you adopted the tool in good faith and the vendor represented it as fair.
• One-employee threshold. A single Illinois employee brings your company within scope. There is no small-employer exemption analogous to Title VII's 15-employee threshold.
• Full lifecycle scope. The law reaches recruitment, hiring, promotion, renewal of employment, training selection, discharge, discipline, tenure, and the terms, privileges, or conditions of employment. AI-driven scheduling, performance management, and productivity monitoring all sit within scope, not just hiring tools.
• Zip code prohibition. Using zip codes as a feature in covered AI decisions is independently prohibited, regardless of intent. This is a documentation question for every vendor in your stack.
• Deployer liability, not vendor liability. The burden sits with you as the deploying employer. Vendor representations are not a defense.

The practical implication: by the time the IDHR finalizes Subpart J in late 2026 or 2027, you should have a documented AI Employment Inventory, a vendor due diligence record, a notice posture, and a monitoring cadence. The IDHR will not be the first audience for that documentation. Plaintiffs' counsel, your D&O underwriter, and your own GC will get to it first.

The HR systems HB 3773 touches

HR leaders typically maintain inventories of HR systems organized by function (recruiting, performance, payroll). HB 3773 cuts across that inventory differently. The relevant question for each system is whether it uses AI to produce or materially influence a covered employment decision — and most modern HR systems now do, in places HR leaders may not have explicitly tracked.

Below is a system-by-system walk-through of where HB 3773 exposure typically lives. Treat this list as a map for building your AI Employment Inventory. The named vendors are illustrative examples of products in each category; this is not a compliance assessment of any specific product, and inclusion does not imply non-compliance.

Applicant Tracking Systems

The major ATS platforms — Workday Recruiting, SuccessFactors, Greenhouse, Lever, iCIMS, Oracle Recruiting Cloud, Taleo — have all integrated AI-driven candidate ranking, screening, and matching features over the past several years. These features may be enabled by default, surfaced through native modules, or accessed through third-party marketplace add-ons.

For each ATS in use, the questions are: which AI features are currently active; what data they are scoring on; how their output influences a recruiter's review or a hiring decision; whether candidate ranking is automated, surfaced as a recommendation, or structured as a filter; and what audit trail exists for AI-influenced candidate dispositions. AI features that are technically optional but operationally relied upon by recruiters are in scope.

Resume screeners and candidate ranking

Tools sold as standalone resume screeners or candidate-ranking services — including some plugins to the major ATSs — are higher-exposure because their entire function is to influence a hiring decision. These products often score on language features, keyword matches, role-fit predictions, or "culture fit" proxies. Each is a potential vector for indirect discrimination on protected characteristics.

Two questions matter most for these tools: what features the model uses (any geographic inputs trigger the zip code prohibition; any inputs correlated with protected-class membership warrant disparate impact testing), and how recruiters use the output. A score presented as a recommendation that recruiters override 40% of the time creates a different evidentiary posture than a score presented as a filter that cuts candidates before recruiter review.

AI video interview platforms

HireVue is the most-named example; Modern Hire, myInterview, and others occupy the same category. Any tool that records candidate video and applies AI analysis — to facial expressions, vocal features, language content, or behavioral signals — sits at the intersection of HB 3773 and the Illinois Artificial Intelligence Video Interview Act (AIVIA, 2019).

AIVIA still applies and adds notice plus consent obligations on top of HB 3773. Where the two overlap, you have to satisfy both. The AIVIA notice must be specifically about the video AI analysis; HB 3773's notice obligation is broader and triggers across the employment lifecycle. A single combined notice can satisfy both, but has to be drafted with both statutes in mind.

Skills assessments and pre-employment testing

Cognitive ability tests, personality inventories, and skills assessments that are scored or interpreted by AI fall in scope when used in a covered employment decision. The major vendors — HireSelect, Pymetrics, Plum, Criteria — all have AI components in either scoring or interpretation, even where the underlying assessment is a more traditional psychometric instrument.

Disparate impact in cognitive and skills assessment tools has been studied and litigated for decades; the AI overlay does not change the underlying Title VII / IHRA analysis but does add the HB 3773 strict-liability question. If the tool is AI-scored, the deployer's burden is documenting the validation evidence the vendor maintains and any independent disparate-impact testing the deployer has performed or commissioned.

AI-driven scheduling and workforce management

Kronos / UKG, ADP Workforce Manager, Quinyx, Legion, and others use AI to generate shift schedules, predict labor demand, and assign workers. These tools sit within HB 3773's scope because shift assignment, weekend rotation, overtime allocation, and similar decisions affect "the terms, privileges, or conditions of employment."

Disparate impact in scheduling AI has been documented — algorithms that optimize for labor cost can systematically assign less desirable shifts to workers in particular demographic groups, particularly when inputs include geographic data, transit-time estimates, or proxies. HR leaders should review what the scheduling AI optimizes for and what features it uses.

Performance management AI

Performance management platforms — Lattice, 15Five, Culture Amp, Workday Performance, Microsoft Viva — increasingly include AI features that summarize feedback, generate review drafts, identify "high-potential" employees, or flag performance concerns. Each feature touches "discipline, tenure, and the terms, privileges, or conditions of employment" within HB 3773's scope.

The exposure is meaningful because performance review outputs drive compensation, promotion, and discharge decisions. AI that systematically scores employees differently across protected classes — through any input feature, intentional or otherwise — produces strict-liability exposure for every downstream personnel action.

Productivity monitoring

Productivity monitoring tools — Teramind, ActivTrak, Hubstaff, Time Doctor, Microsoft Productivity Score — produce data that feeds into discipline and discharge decisions. When the monitoring uses AI to characterize productivity, identify "underperforming" employees, or flag anomalies, the AI is making the covered employment decision in fact even if a human supervisor signs the discipline letter.

The compliance question is the same: what features the AI uses, how its output influences the human decision, and what evidence exists of disparate impact across protected classes.

L&D recommendation engines

Learning & development recommendation engines — Cornerstone, Degreed, EdCast, LinkedIn Learning — recommend training and development opportunities to employees. HB 3773 covers "selection for training or apprenticeship." Recommendation engines that systematically surface different opportunities to different demographic groups are in scope.

L&D AI is lower-exposure than hiring or performance AI, but the structure of the inquiry is the same: what features the recommendation model uses, how recommendations are surfaced to employees, and what evidence exists that the surfacing patterns are not correlated with protected-class membership.

The 6-question vendor evaluation

The fastest way to build defensible vendor due diligence is to send every HR-AI vendor in your stack the same six questions and require a written response. The responses become your due-diligence file. Vendor responses also surface which vendors are mature enough to operate in a strict-liability jurisdiction and which are not.

The six questions:

1. What features does the AI use?

Specifically: does the model use, as an input or training signal, any of the following — zip code, home address, neighborhood, school attended, name, native language, photograph, video, voice recording, age, or any feature that could correlate with race, ethnicity, gender, national origin, disability, age, sexual orientation, gender identity, marital status, military status, or pregnancy. Any "yes" response triggers either an immediate compliance question (the zip code prohibition is express) or a disparate-impact testing question.

2. What testing has the vendor performed?

For the Illinois deployment specifically, has the vendor tested for disparate impact across IHRA-protected classes? What test methodology, what population, what results, what remediation? "We have a fairness team" is not an answer. The deliverable is a written summary of testing on a representative population, with methodology and results.

3. Where does the vendor draw the line between AI recommendation and AI decision?

Permitted use under IDHR draft Subpart J turns in part on whether AI "influences" or "facilitates" a decision versus making the decision outright. The vendor should be able to describe, in workflow terms, where the human decision-maker sits relative to the AI output and what evidence exists that the human is in fact exercising independent judgment.

4. What recordkeeping does the vendor maintain?

What records does the vendor preserve for each AI-mediated decision — input features, model version, output score, confidence intervals, recommendation strength, and downstream decision? For how long? How is that data made available to you on request, both routinely and in response to an IDHR charge or civil discovery request?

5. How does the vendor handle model updates?

Modern AI tools update continuously. A model that tested well for disparate impact in March may behave differently in August. How does the vendor notify deployers of material model updates? How often is the disparate-impact testing repeated? What is the rollback path if a deployment surfaces unexpected impact?

6. What are the vendor's contractual representations?

Specifically: does the vendor contractually represent that the product complies with HB 3773 and applicable Illinois employment law; does the vendor indemnify the deployer against compliance failures; and what is the termination right if the vendor's product is implicated in an enforcement action or private suit? The answers do not change your strict-liability exposure (vendor representations are not a defense) but they do affect economic allocation of compliance risk.

Send these six to every HR-AI vendor in the stack and give them a response deadline. The responses build your file. Refusals or non-responses are themselves diligence findings.

The notice question

HB 3773 requires employers to provide notice to employees when AI is used in a covered employment decision. The statute itself does not specify the timing, form, or content of the notice; those details are delegated to IDHR through Subpart J rulemaking.

As of May 2026, draft Subpart J has been circulated to stakeholders but not formally published for public comment. The draft framework contemplates notice at or before the point of an AI-mediated decision, in writing, with content sufficient to inform the employee that AI is being used, the categories of AI use, and how the employee can obtain more information.

HR leaders cannot wait for final rules to draft notice. The operative approach is to build notice templates against the draft framework now, deploy them, and refresh when final rules publish. The structural elements unlikely to change between draft and final:

• Notice must be in writing
• Notice must precede or accompany the AI-mediated decision
• Notice must identify that AI is being used (without disclosing trade-secret model details)
• Notice must describe the categories of AI use relevant to the recipient
• Notice must inform the recipient how to obtain additional information

Where to deploy the notice depends on the AI touchpoint. Pre-hire AI (resume screening, video interview, assessments) typically warrants notice in the job posting, application portal, and interview confirmation. Post-hire AI (performance management, scheduling, productivity monitoring) typically warrants notice in the offer letter, employee handbook, internal portal, and at deployment of any new AI tool.

Build the notice infrastructure inside existing employee communication channels — applicant tracking system, offer letter templates, handbook, employee portal — rather than as standalone documents. The compliance posture is "notice was given through the same channel by which employment communications normally reach the employee, contemporaneous with the AI-mediated decision."

The IDHR AI Rulemaking Tracker documents draft Subpart J in detail and tracks formal publication progress.

What to tell your CEO

HR leaders need an exec-conversation framework for HB 3773. The pattern that works: a one-page brief covering five questions execs consistently ask, with a position on each.

"What's our exposure today?"

Strict liability for any AI tool in the employment lifecycle that produces a discriminatory effect on a protected class. The exposure includes private civil action by any employee or applicant, IDHR charge of civil rights violation, class action treatment for systematic AI-tool exposure, and reputational exposure if an enforcement action becomes public. The exposure exists now — January 1, 2026 — regardless of whether IDHR has finalized implementing rules.

"What's the cost of compliance?"

Three tiers. Documentation-only posture — AI Employment Inventory, vendor due diligence, notice templates, posture memo: 4–8 weeks of HR plus IT time, supplemented by external advisory engagement. Documentation plus targeted disparate-impact testing on highest-risk tools: another 2–4 weeks of consulting and statistical work. Full remediation if testing surfaces problems: scope-dependent on vendor remediation, replacement, or process redesign. The avoided cost — a single strict-liability class action against a widely-used HR AI tool — is materially larger than any of the three tiers.

"Are we ahead of or behind peers?"

The honest answer for most Illinois employers in mid-2026: behind. The widespread industry pattern is to wait for IDHR final rules before building compliance posture, despite the statute being operative. Companies that build documented posture before the IDHR finalizes will be insulated when the first wave of enforcement and private litigation hits. Companies that wait will be exposed in proportion to their AI-tool footprint.

"Can we just rely on our vendors?"

No. The strict-liability standard places the compliance burden on the employer-deployer. Vendor representations create a due-diligence record but are not a defense. The vendor accountability question is contractual — what indemnities the vendor is willing to write into the agreement — not a substitute for the deployer's compliance posture.

"What's our timeline?"

The 90-day calendar below. Days 0–30 are inventory and vendor outreach. Days 30–60 are notice infrastructure and high-risk testing. Days 60–90 are documentation and exec readout. After 90 days, the posture is maintained through quarterly review and triggered refresh on any new AI procurement or vendor model update.

The 90-day HR compliance calendar

The full compliance posture is buildable in 90 days for most mid-market Illinois employers. The sequence:

Days 0–30 — Inventory and vendor outreach

Week 1: Build the AI Employment Inventory. List every HR system in use, every AI feature within those systems (whether enabled by default or activated separately), and the covered employment decisions each feature touches. The inventory typically surfaces AI features that HR leaders did not realize were active.

Week 2: Send the 6-question vendor evaluation to every HR-AI vendor in the stack. Give vendors a 14-day response window. Track which vendors respond and which do not.

Week 3: Convene the cross-functional working group — HR, IT, Procurement, GC or outside counsel, Privacy if separate. Agree on the scope of the compliance project, the project owner, the decision rights, and the escalation path. Document the working group's charter.

Week 4: Review the vendor responses received. Identify the top 3–5 highest-exposure tools based on (a) the volume of covered decisions they influence, (b) the breadth of features they use, and (c) the quality of the vendor's due-diligence response. These will be the targets for disparate-impact testing.

Days 30–60 — Notice infrastructure and high-risk testing

Week 5: Draft notice templates against the draft Subpart J framework. Build pre-hire notice into application portals and interview confirmations. Build post-hire notice into the employee handbook, the offer letter template, and internal communications for any AI tool deployment.

Week 6: Initiate disparate-impact testing on the highest-risk tools. This may be done internally if you have data science capacity, by the vendor under your supervision, or by an independent consultant. The test scope: each protected class under the IHRA, on your Illinois applicant or workforce data, against the AI tool's output.

Week 7: Review testing results. For any tool with documented disparate impact, document remediation options — vendor remediation, process change, tool replacement — and decision criteria. Begin vendor remediation conversations where applicable.

Week 8: Update procurement processes. Add HB 3773 compliance review to the vendor onboarding workflow. No new HR-AI tool should be procured without running through the 6-question evaluation and being recorded in the inventory.

Days 60–90 — Documentation and exec readout

Week 9: Draft the compliance posture memo. The 5–10-page memo summarizes the AI Employment Inventory, the vendor due diligence record, the notice infrastructure, the disparate-impact testing results, and any remediation actions taken or planned. The memo is the single most important defensive document.

Week 10: Internal review of the memo with GC, the working group, and senior HR leadership. Refine.

Week 11: Brief the CEO and, where relevant, the audit or risk committee of the board. Use the exec-conversation framework above. The brief positions HR's work as risk reduction within an operative legal regime — not as a compliance bureaucracy.

Week 12: Establish the quarterly maintenance cadence. The AI Employment Inventory refreshes quarterly. New AI procurement runs through the vendor evaluation. Disparate-impact testing repeats annually for the highest-risk tools, or more frequently after material vendor model updates. The compliance posture memo refreshes annually or on any material change in the AI footprint.

When to escalate beyond HR

HR owns the operational compliance posture, but several scenarios require escalation outside HR. Escalate to GC or outside counsel when:

• Disparate-impact testing surfaces material findings warranting legal review of remediation options
• An employee or applicant raises an HB 3773 concern, files an internal complaint, or files an IDHR charge
• A vendor refuses to respond to the 6-question evaluation or provides responses inconsistent with strict-liability compliance
• The notice question intersects with collective bargaining agreement terms
• The company's D&O renewal cycle is approaching and the compliance posture needs to be documented for the carrier

Escalate to the board (or its audit / risk committee) when:

• The compliance posture is established and HR is briefing the board as part of routine oversight (recommend the first compliance posture briefing within 12 months of January 1, 2026)
• A material incident occurs — AI tool failure, regulatory inquiry, private suit, or significant disparate-impact finding
• A material AI footprint change is contemplated — major new vendor, replacement of a primary HR system, expansion into new Illinois jurisdictions

The board-level analysis sits within fiduciary-duty doctrine and D&O liability questions, covered in the AI Risk and D&O Liability reference. HR's role is to produce the documentation that the board and the D&O carrier need; the legal analysis lives with counsel.

Employers operating across multiple states beyond Illinois should also see the Multi-Jurisdictional AI Compliance Framework, which describes how the Illinois posture overlays the Colorado AI Act, NYC Local Law 144, California ADMT regulations, the EU AI Act, and adjacent frameworks.

---

This article was last reviewed on May 20, 2026. As IDHR Subpart J rulemaking progresses and case law develops, the article will be updated. For HR leaders who want a printable working document, the HR Leader's HB 3773 Compliance Checklist collects the 6-question vendor evaluation, the notice scaffolding, and the 90-day calendar in a single 6-page working PDF. For HR leaders who want movement this quarter without a full compliance review, the HB 3773 HR Workshop is a 2-session fixed-fee engagement that produces a custom inventory, plan, and risk advisory in roughly four working weeks. For the full scope, see the HB 3773 Employment AI Compliance Review. For the underlying legal framework and the IDHR rulemaking status, see the HB 3773 Compliance Guide and the IDHR AI Rulemaking Tracker. The Illinois AI Legislative Ecosystem tracker at strategy.techne.ai maintains real-time tracking of rulemaking and enforcement developments.