IDHR AI Rulemaking Tracker: Subpart J and HB 3773 Implementation

Current rulemaking status

As of May 9, 2026: IDHR has shared draft Subpart J with stakeholders and the National Federation of Independent Business reported on the substance of the draft in late 2025. Formal publication in the Illinois Register has not yet occurred. IDHR is reportedly continuing to refine the draft in response to stakeholder feedback.

Operative statute: HB 3773 (Public Act 103-0804), effective January 1, 2026. The statutory text governs employer obligations until and unless final implementing rules are adopted.

Federal context: The December 2025 federal Executive Order on AI policy directs the U.S. Department of Justice to evaluate state AI laws for federal preemption challenges. No preemption challenge has been filed against HB 3773 specifically as of the article date.

This section will be updated as IDHR formally publishes Subpart J, holds public hearings, advances through JCAR review, or adopts final rules. Last update: May 9, 2026.

Statutory authority

HB 3773 (Public Act 103-0804) directs IDHR to adopt "any rules necessary for the implementation and enforcement" of the statute's AI provisions. The statute specifies that rulemaking shall include, but not be limited to, "rules on the circumstances and conditions that require notice, the time period for providing notice, and the means for providing notice."

The rulemaking must follow the Illinois Administrative Procedure Act (5 ILCS 100), which requires:

• Publication of proposed rules in the Illinois Register
• A 45-day public comment period
• JCAR (Joint Committee on Administrative Rules) review
• Publication of final rules with response to comments

IDHR's draft Subpart J — circulated in stakeholder consultation but not yet in the Illinois Register — represents the agency's initial framework before formal publication. Past IDHR rulemaking cadence suggests stakeholder consultation can refine drafts substantially before formal publication, but the core architecture rarely changes.

What draft Subpart J says

Based on stakeholder reporting and analysis published by employment law practitioners in late 2025 and early 2026, draft Subpart J addresses the following components:

Scope of "covered employment decision"

Draft Subpart J defines "covered employment decision" broadly, consistent with the statutory language. It includes decisions with respect to:

• Recruitment
• Hiring
• Promotion
• Renewal of employment
• Selection for training or apprenticeship
• Discharge
• Discipline
• Tenure
• Terms, privileges, or conditions of employment

This is functionally the full life cycle of the employment relationship. AI use that influences any decision in this list triggers notice obligations and falls within the scope of the discrimination prohibition.

The "influence or facilitate" standard

Notice is required whenever an employer uses AI to "influence or facilitate" any covered employment decision. This is a deliberately broad standard. AI does not need to be the dispositive decision-maker; it need only influence or facilitate the decision-making process. Resume screening, candidate ranking, interview scheduling, performance scoring, and predictive analytics used in any aspect of employment decision-making fall within scope.

Notice form and timing

Draft Subpart J anticipates that notice must be provided:

• Initially, before AI is used to make or influence a covered employment decision affecting an applicant or employee
• Annually thereafter
• Whenever a new or substantially updated AI tool is introduced into a covered employment workflow

Notice must be posted in physical and online locations accessible to employees and applicants. Specific format requirements (e.g., minimum content, font size, language accessibility) are expected to be specified in final rules.

Notice failure as an independent violation

Critically, draft Subpart J makes an employer's failure to provide notice an independent violation of the IHRA, separate from any discrimination claim. Notice is not merely procedural; the absence of notice is itself actionable. This means an employer can violate HB 3773 even if its AI use is fully non-discriminatory, simply by failing to provide proper notice.

Recordkeeping

Draft Subpart J requires employers to maintain records of AI use in covered employment decisions, including documentation of the AI tools used, the workflows in which AI was applied, and the notices provided. Specific retention periods are expected to align with other IHRA recordkeeping obligations (typically one to three years).

Application to employer agents

The draft applies broadly to "employers" under the IHRA and to their agents — including recruiters, third-party AI vendors, and others acting on the employer's behalf. The allocation of compliance obligations between employer and agent is one of the more complex open questions discussed below.

Stakeholder feedback themes

Based on practitioner reporting and trade association statements, the following themes have emerged in stakeholder feedback to IDHR:

1. Concerns about notice timing and form

Employers and trade associations have raised concerns about the practical mechanics of providing initial notice "before" AI use in every recruitment workflow. The concern is particularly acute for high-volume hiring and for AI tools used early in the recruitment funnel (e.g., resume parsers running in the background of an applicant tracking system). Stakeholders are pushing for clearer guidance on what constitutes adequate "advance" notice and whether posted notice (vs. individualized notice to each applicant) is sufficient.

2. Vendor liability allocation

Employer-side stakeholders are concerned about being held responsible for the AI behavior of third-party vendors over which they have limited visibility. Vendor-side stakeholders are pushing back against obligations being imposed directly on them. The draft's broad application to "agents" leaves this allocation unclear, and final rules will likely need to specify how the obligations apply across the employer-vendor relationship.

3. Definition of "AI" for purposes of the rule

The statutory definition of "Artificial Intelligence" in HB 3773 is broad — "a machine-based system that, for explicit or implicit objectives, infers... how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments." Stakeholders have asked whether this captures basic statistical scoring, rule-based screening tools that do not employ machine learning, and traditional ATS keyword filtering. IDHR has signaled an intent to interpret the definition broadly, consistent with the statutory text.

4. Coordination with multi-state operations

Employers operating across multiple states have asked for guidance on how Illinois requirements interact with NYC Local Law 144 (bias audits), Colorado AI Act (impact assessments), and California ADMT (pre-use notice and opt-out). IDHR has not committed to specific coordination provisions but has signaled awareness of the multi-jurisdictional reality.

5. Recordkeeping burden

Small employers and trade associations have raised concerns about the cumulative recordkeeping burden of HB 3773 alongside existing IHRA, FCRA, and federal employment recordkeeping requirements. IDHR has indicated that recordkeeping requirements will be calibrated to avoid duplication where possible.

Open questions and ambiguities

Several substantive questions remain open as final rules are developed:

1. Are vendor representations a defense?

If an employer relies on vendor representations that an AI tool is non-discriminatory, and the tool is later found to produce discriminatory outcomes, can the employer defend itself by pointing to vendor representations? HB 3773 imposes strict liability — intent does not matter — so vendor representations alone are unlikely to be a complete defense. But the relationship between vendor due diligence and employer liability is one of the most practically important questions.

2. How do strict-liability claims interact with disparate-impact precedent?

The IHRA already prohibits practices with disparate impact on protected classes. HB 3773's strict liability provision adds an additional layer — AI use that has the effect of subjecting employees to discrimination is prohibited. The relationship between these two standards (whether HB 3773 is a separate cause of action or a sharpening of the disparate-impact framework) is not yet clear.

3. What constitutes an "AI tool" being introduced or substantially updated?

Notice is required when a new or substantially updated AI tool is introduced. But AI vendors continuously update their models. Does every model update trigger renewed notice? Is there a materiality threshold? Final rules will need to address this.

4. Does notice need to be individualized or posted?

The draft contemplates posting notice in physical and online locations, but it is unclear whether posted notice is sufficient or whether individualized notice to each applicant or employee is required. Posted notice is operationally simpler; individualized notice may be required by parallel state laws (California ADMT) and is the safer compliance posture.

5. How is "agent" defined for purposes of derivative liability?

The draft applies to employers' agents but does not define the boundary. Are recruitment platforms agents? Are background check providers agents? Are SaaS HR vendors agents? The breadth of agent liability is one of the most consequential open questions for AI vendors operating in Illinois.

Federal preemption considerations

On December 17, 2025, the federal government issued an Executive Order titled "Ensuring a National Policy Framework for Artificial Intelligence." Among other provisions, the EO directs the U.S. Attorney General to establish an AI Litigation Task Force to evaluate state AI laws and challenge those deemed inconsistent with federal policy.

HB 3773 has been mentioned in commentary as a possible target for a preemption challenge, on theories that include:

• Conflict preemption with the federal AI policy framework
• Commerce Clause implications for AI vendors operating across state lines
• Section 230 implications for AI platforms
• First Amendment implications for AI as expressive technology

As of May 2026, no preemption challenge has been filed against HB 3773. Employers preparing compliance posture should not assume preemption will eliminate their obligations; the more prudent posture is to comply with HB 3773 while monitoring for federal developments.

Comparison to other state rulemaking

IDHR's rulemaking approach can be situated against parallel rulemaking in adjacent jurisdictions:

NYC Local Law 144

The NYC Department of Consumer and Worker Protection (DCWP) rulemaking for LL 144 — automated employment decision tools — focused on bias audit requirements, candidate notice, and audit publication. IDHR has not adopted a parallel bias audit requirement. IDHR's emphasis is on notice and recordkeeping rather than substantive testing.

Colorado AI Act

The Colorado Department of Labor and Employment (CDLE) and Colorado Department of Regulatory Agencies (DORA) rulemaking under the Colorado AI Act focuses on impact assessments, deployer obligations, and risk management programs. The Colorado approach is more prescriptive than draft Subpart J.

California ADMT regulations

The California Privacy Protection Agency (CPPA) has finalized regulations governing automated decision-making technology under the CCPA. The California approach focuses on pre-use notice, opt-out rights, and access rights — a privacy-anchored framework that differs structurally from Illinois's antidiscrimination approach.

Multi-state employers face the practical challenge of complying with all four frameworks simultaneously. A unified compliance program built to the most rigorous applicable standard is generally more efficient than maintaining separate jurisdiction-specific programs.

What employers should do during the rulemaking period

Waiting for final rules before acting is not a viable compliance posture. Employers should take the following steps now:

1. Build the AI inventory

Identify every AI tool used in any aspect of recruitment, hiring, promotion, performance evaluation, or any other covered employment decision. Include both first-party tools and tools embedded in third-party platforms (ATS, ATS vendors, sourcing platforms, interview platforms).

2. Draft notice templates against draft Subpart J

Build notice templates that satisfy the apparent draft requirements: initial notice before AI use, annual posted notice, notice on introduction of new tools. Treat the draft as a strong indicator of final form. Templates can be finalized once final rules are published.

3. Update the AI vendor due diligence framework

For each AI tool in the inventory, document the vendor's representations about non-discrimination, the testing the vendor has performed, and the AI tool's role in the employment decision workflow. The records will be useful both for HB 3773 compliance and for D&O renewal documentation.

4. Establish recordkeeping practices

Document the AI tools used, the dates of introduction or substantial update, the notices provided, and the workflows in which AI was applied. The recordkeeping practice should be operational by the time HB 3773 takes effect (already January 1, 2026), regardless of final rule timing.

5. Engage with the rulemaking process

When IDHR formally publishes Subpart J for public comment, submit substantive comments. The comment period is the formal mechanism for influencing final rule content. Trade association coordination can amplify employer perspectives.

Timeline projection

Based on past IDHR rulemaking cadence and the volume of stakeholder feedback received to date:

Stage	Estimated timing
Stakeholder consultation continues	Ongoing through Q2 2026
Formal publication in Illinois Register	Q3-Q4 2026
45-day public comment period	Q4 2026 - Q1 2027
JCAR review	Q1-Q2 2027
Final rule adoption and publication	Q2-Q3 2027

These projections are subject to revision based on federal preemption developments, the volume and substance of public comments, and IDHR's prioritization. Until final rules are adopted, the statutory text of HB 3773 is the operative law.

---

This article is updated monthly as IDHR rulemaking progresses. For the most recent status, see the date stamp at the top. HR leaders who own operational compliance should see the Operating Playbook for HR Leaders, which translates the statute and the draft Subpart J framework into vendor evaluation, notice scaffolding, and a 90-day compliance calendar. Employers needing to develop a broader compliance posture should also see the HB 3773 Compliance Guide and the Illinois AI Legislative Ecosystem tracker. Employers operating across multiple jurisdictions should see the Multi-Jurisdictional AI Compliance Framework, which describes how Illinois requirements layer onto the EU AI Act, the Colorado AI Act, NYC Local Law 144, and California ADMT. Behavioral health operators in Illinois should additionally see the Illinois WOPR Act Compliance Reference. Boards evaluating the fiduciary implications of HB 3773 strict liability exposure should see AI Risk and D&O Liability for Directors. For an organization-specific HB 3773 compliance review, see the HB 3773 Employment AI Compliance Review service.