What Illinois Employers Must Disclose About Hiring AI Now
HB 3773 is in effect, but Illinois withdrew its detailed AI notice rules. See which hiring tools are implicated and build a defensible interim notice.
Illinois employers should treat HB 3773’s AI notice duty as operative now. The employment-AI amendments to the Illinois Human Rights Act (IHRA) took effect January 1, 2026, and the cited legal analyses report that withdrawal of the implementing rules did not suspend the statute. Resume ranking, targeted recruiting, interview scoring, promotion recommendations, and similar uses can raise notice issues when AI influences a covered employment decision. The exact wording, timing, delivery, and record-retention mechanics remain unsettled.
The Illinois Department of Human Rights (IDHR) published proposed rules on May 15, 2026, with comments scheduled through June 29. It then temporarily withdrew or postponed the proposal to coordinate with other state agencies. Employers should use the withdrawn provisions as planning guidance, not binding law, while verifying the current IHRA, IDHR docket, and any agency action after the developments covered by the cited June analyses.
Select every AI use in your workflow; the builder will produce an interim notice, delivery plan, and separate video-interview warning.
Illinois Interim Notice Builder
Build a Use-Specific AI Notice Plan
Choose the functions actually used for Illinois applicants or employees. The default example shows resume screening. Proposed-rule details are labeled as planning guidance rather than final requirements.
1. Select Every AI-Assisted Function
This is an issue-spotting tool, not a determination of statutory coverage.
2. Generated Interim Notice
Draft Notice Text
Fields to Complete Before Publishing
- Employer’s legal or commonly used name
- Product and vendor or developer
- Positions, departments, or groups affected
- Employment purpose and decision stages
- Broad categories of data processed
- Output produced and how reviewers use it
- HR or compliance contact
- Accommodation or alternative-process channel
Proposal-informedPlain languageVerify current law
3. Delivery and Timing Plan
Evidence to Preserve
- Notice version and approval record
- Publication or delivery date and channel
- Recipient group and positions covered
- Product, feature, configuration, and vendor information
- Adoption date and material-change history
- Accommodation and inquiry routing process
The regulatory history described three years in the May proposal and four years in earlier draft reporting. Neither is presented here as a binding retention period; use a counsel-approved records schedule.
4. Function-by-Function Decision Matrix
| Function | Interim Concern | Decision Connection | Notice Detail to Capture |
|---|---|---|---|
| Generic job-description drafting | Lower | No person analyzed and no outcome recommended | Document the content-only limitation and reassess if the workflow changes |
| Targeted recruitment advertising | Higher | Influences who receives access to an opportunity | Audience data, inferred groups, optimization purpose, positions, and campaign controller |
| Resume parsing or summarization | Context-Dependent | May omit, prioritize, compare, or characterize qualifications | Inputs, output type, affected roles, and how recruiters rely on the output |
| Candidate ranking | Higher | Can alter review order, advancement, or practical consideration | Ranking purpose, criteria or data categories, positions, and human-review process |
| Generic interview-question drafting | Lower | Does not itself analyze a candidate or score an answer | Document that candidate data and decision recommendations are excluded |
| Interview answer scoring | Higher | Can affect selection, rejection, or advancement | Answer data, score or recommendation, affected roles, and reviewer use |
| Recorded applicant video analysis | Higher + Separate Law | Analyzes an applicant-submitted recording | IHRA notice analysis plus advance explanation, consent, and deletion process under the separate act |
| Employee performance summary | Context-Dependent | May influence compensation, promotion, discipline, or discharge | Data sources, characterization produced, affected employees, and management use |
| Promotion or discipline recommendation | Higher | Directly concerns a reported covered employment purpose | Recommendation type, employee groups, decision stage, and human-review process |
| Work assignment or scheduling | Context-Dependent | May affect terms, privileges, opportunities, or conditions of employment | Scheduling inputs, optimization goal, affected workforce, and override process |
Source note: HB 3773/IHRA status and proposed-rule details are drawn from the cited June 2026 legal analyses in the article. The May IDHR proposal was reported withdrawn or postponed; its product, content, posting, 30-day, annual, and retention details are not presented as final law.
The Statutory Duty Survived Withdrawal of the Rules
HB 3773 amended the IHRA, and its employment-AI provisions became effective on January 1, 2026. Secondary legal analyses generally describe the amendments as requiring notice to applicants or employees when AI is used for specified employment purposes. Employers should confirm the precise recipients and operative wording against current primary Illinois authority.
IDHR’s May proposal would have supplied much more detailed instructions. In June, legal updates described the agency’s action as a withdrawal, temporary withdrawal, postponement, or pause. They nevertheless reported that the statutory obligations remained effective. Seyfarth Shaw’s analysis describes the effective date, June 2 postponement, and continuing statutory obligations.
The available evidence therefore supports this division:
| Status | Employer Treatment |
|---|---|
| Reported statutory notice duty | Treat it as operative while confirming employer coverage, recipients, location, and the use involved. |
| Withdrawn IDHR proposal | Use it as regulatory history and a conservative design reference, not binding law. |
| Artificial Intelligence Video Interview Act | Conduct a separate notice, explanation, consent, and deletion analysis for covered recorded interviews. |
| Interim controls | Inventory tools, issue accurate communications, preserve delivery evidence, and prepare to revise the process. |
The supplied evidence does not establish final statewide rules for exact notice wording, individualized delivery, acknowledgment, translation, accessibility procedures, or retention. It also does not establish whether IDHR took additional action after the cited June updates. Current primary sources and qualified Illinois counsel should be checked before relying on a compliance position.
The Regulatory Timeline Explains the Present Gap
Governor J.B. Pritzker signed HB 3773 on August 9, 2024. Before the amendments took effect, IDHR circulated draft language addressing definitions, notice procedures, posting, accessibility, and records.
The employment-AI amendments took effect January 1, 2026. IDHR published proposed amendments on May 15, covering notice content and delivery in greater detail. According to June legal updates, the agency announced a temporary withdrawal or postponement on June 2 to continue coordination with other state agencies. The hearing scheduled for June 10 was canceled.
Some articles were updated on June 10 to report that development, which explains why accounts use either June 2 or June 10. The cited updates did not identify a replacement schedule. Kilpatrick Townsend reports the May publication, temporary withdrawal, and continued effect of the statute.
The withdrawal was not reported as a repeal, delayed effective date, or suspension of HB 3773. It removed the proposed implementation details from the active rulemaking process described by those sources.
Notice Turns on the Tool’s Function and Employment Influence
Legal summaries report that the amendments address AI used for recruitment, hiring, promotion, renewal of employment, selection for training or apprenticeship, discharge, discipline, tenure, and other terms, privileges, or conditions of employment.
That scope reaches beyond automated rejection. A system can influence a decision by ranking candidates, deciding which people receive recruitment advertising, summarizing qualifications, scoring interviews, recommending training, classifying performance, allocating work, or flagging employees for discipline.
Product labels are not reliable compliance categories. An applicant tracking system may be a database in one workflow and a ranking engine in another. A general-purpose AI assistant may draft generic text or analyze a particular applicant. Review each feature and use separately.
| Use | Interim Notice Assessment | Reason |
|---|---|---|
| Drafting a generic job description without analyzing people | Lower concern | It is less connected to an individual employment decision. The withdrawn proposal’s treatment of content-only uses is not a binding exemption. |
| Targeting recruitment advertising to inferred groups | Higher concern | AI may determine who receives practical access to an employment opportunity. |
| Parsing or summarizing resumes | Context-dependent | Concern increases when the output omits, prioritizes, compares, or characterizes qualifications for a reviewer. |
| Ranking applicants | Higher concern | Ranking can affect review order, advancement, or access to consideration. |
| Generating generic interview questions | Lower concern | The tool is less likely to implicate notice if it does not analyze candidate data or recommend outcomes. |
| Scoring interview answers | Higher concern | A score can directly influence selection, rejection, or advancement. |
| Analyzing a recorded applicant interview | Higher concern plus separate law | The use may implicate both the IHRA framework and the Artificial Intelligence Video Interview Act. |
| Summarizing employee performance | Context-dependent | A clerical summary differs from an output that characterizes performance for compensation, promotion, or discipline. |
| Recommending promotion, discipline, or discharge | Higher concern | The output directly concerns reported categories of covered employment decisions. |
A human making the final choice does not necessarily remove the use from the analysis. AI still influences a decision when a recruiter sees a rank, an interviewer receives an automated score, or a manager relies on an AI-generated performance classification.
For each use, determine whether the system processes applicant or employee information; infers, classifies, predicts, scores, summarizes, recommends, or decides; and can change who is sourced, reviewed, interviewed, advanced, hired, trained, scheduled, promoted, disciplined, or discharged.
Employers should also inspect embedded and informally adopted features. Record which features are enabled, whether outputs appear automatically, who uses them, and whether the vendor can change models or data sources. Include recruiter browser extensions and unapproved general-purpose AI, not just centrally purchased platforms.
Resume Screening and Targeted Recruiting Commonly Raise Notice Issues
Resume parsing, screening, summarization, and ranking require a feature-level review. Simple extraction of contact details is different from scoring qualifications or recommending which candidates a recruiter should open first. If the recruiter’s process would materially change without the output, the use belongs in the notice analysis.
Targeted recruiting deserves the same attention even though it occurs before an application. A system that predicts who should receive a job advertisement can influence recruitment by controlling access to the opportunity. Employers should identify the audience data, inferred traits, optimization objective, affected positions, and party controlling the campaign.
External recruiters, staffing firms, recruitment-process outsourcers, assessment providers, and vendors can complicate responsibility. Contracts should identify who selects and configures the tool, supplies data, receives its output, provides notice, preserves delivery evidence, handles accommodation requests, and reports product changes.
A contract can assign operational responsibilities and indemnification. It does not necessarily determine which entity has statutory responsibility under the IHRA.
The Withdrawn Proposal Supplies a Useful Notice Blueprint
The May proposal reportedly addressed product and vendor identification, affected decisions, purpose, data categories, positions, contact information, accommodation rights, posting locations, timing, translation, accessibility, and records. Hinshaw’s analysis is useful for the proposed mechanics and separate video-interview rules, but not as proof that those mechanics became final.
A defensible interim notice can therefore identify:
- The employer using the system
- The product, developer, or vendor
- The employment purpose and decision stages affected
- The positions, departments, or groups affected
- Broad categories of applicant or employee data processed
- The score, summary, classification, recommendation, or other output produced
- How that output supports or influences human review
- An HR or compliance contact
- A channel for requesting an accommodation or alternative process
These are proposal-informed fields, not a declaration that every field is currently mandatory. The notice should describe actual configurations rather than every capability advertised by the vendor.
A practical core statement is:
[Employer] may use [tool and vendor] to assist with [employment purpose] for [positions or groups]. The system may process [data categories] and produce [output]. That output may inform [identified decision]. Questions may be directed to [contact]. To request an accommodation or alternative process, use [channel].
Avoid a statement that AI “may be used anywhere” when more accurate detail is available. Also avoid language so narrow that a feature update immediately makes it false. Modular sections let the employer update one system or workflow without replacing the entire policy.
Posting and Timing Remain Proposal-Informed Rather Than Final
The supplied evidence does not establish a universal requirement to place an AI disclosure in every job posting. Job-posting notice for prospective employees appeared in the withdrawn proposal.
The proposal also reportedly contemplated annual employee notice and notice within 30 days after adopting or substantially updating AI. Those intervals are useful planning triggers, but the evidence does not establish them as current statewide deadlines.
Delivery should match the recipient and workflow. A job posting or application portal can reach candidates. Direct email can support individualized delivery. A handbook, employee policy, intranet, or workplace posting may suit post-hire systems. An external AI-use page can provide detail but may not reach a particular person by itself.
A layered approach may be more defensible than a buried general statement. Preserve the notice version, publication or delivery date, recipient group, covered positions, delivery channel, approval record, and related system inventory.
The regulatory history does not supply a reliable binding retention period. Sources described three years in the May proposal and four years in earlier draft reporting. Neither period should be presented as the current statewide requirement. Use a counsel-approved records schedule that accounts for employment-record obligations and legal holds.
Recorded Video Analysis Requires a Separate Consent Review
The Illinois Artificial Intelligence Video Interview Act applies when an employer asks an applicant to record an interview and uses AI to analyze the applicant-submitted recording.
Before a covered interview, the employer must notify the applicant that AI may analyze the video, explain how the AI works and the general types of characteristics it evaluates, and obtain consent. The law also provides a deletion process when an applicant requests deletion; employers should consult the current statute for its exact procedure and timing.
A general HB 3773 notice may not satisfy those requirements. Conversely, Video Interview Act consent should not automatically be extended as a legal requirement to every resume screener, chatbot, live interview, transcript, assessment, or audio-only tool.
For recorded-video workflows, document whether the employer requests the recording, whether AI analyzes it, when notice and explanation are delivered, how consent is captured, and who handles deletion requests. Provide an alternative process where required or appropriate.
A Defensible Interim Program Starts With a Use-Level Inventory
Inventory AI across sourcing, advertising, application intake, assessments, interview tools, background processes, work assignment, productivity monitoring, training selection, promotion and compensation, discipline, and termination.
For every function, record the business owner, vendor, product, enabled feature, users, inputs, data sources, outputs, Illinois positions and people affected, employment stage, degree of influence, adoption date, and last material change. “Vendor X is used in recruiting” is not enough; the record should state what a particular feature does for identified positions.
Assign responsibility across HR, legal, procurement, IT, accessibility teams, hiring managers, staffing partners, and vendors. Vendor contracts and internal controls should require notice of replaced models, new features, material functionality changes, new data sources, scoring changes, integrations, and features enabled by default.
Name an owner to monitor IDHR materials and replacement rulemaking. The process should be capable of changing recipients, timing, content, channels, accessibility measures, and retention without being rebuilt.
Notice Does Not Cure Discriminatory AI Use
The cited analyses report that the IHRA amendments prohibit employment uses of AI that result in discrimination based on classes protected under Illinois law, including unintentional discriminatory effects. They also report a restriction involving ZIP codes used as proxies for protected characteristics. These propositions should be confirmed against the current IHRA.
The supplied evidence does not establish a general HB 3773 mandate for a formal bias audit or AI impact assessment. Employers can still test outputs before deployment and after material changes, review selection patterns, examine data sources for geographic or historical disparities, investigate vendor claims, define override procedures, and document meaningful human review.
Notice is communication, not authorization. An accurate disclosure does not excuse discrimination, a defective accommodation process, a privacy violation, or failure to comply with the separate recorded-video law.