Skip to main content

AI governance

AI Used by Recruiting Agencies: An Illinois Employer Diligence Reference

5 min read
On this page

Summary

A practical guide to documenting AI used through recruiters, staffing firms and other employment suppliers. It distinguishes the enacted Illinois AI provision from fact-specific questions about agency, contractual responsibility and the records needed to understand the decision chain.

Start with the relationship and the decision

A recruiting agency can influence an employer’s hiring process without the employer directly licensing the agency’s software. That is a reason to ask how the service works. It is not sufficient, by itself, to determine the legal responsibility of every party.

Illinois HB 3773 amended section 2-102 of the Human Rights Act to address covered employer AI use, discriminatory effects, ZIP-code proxies and notice. Whether a particular intermediary’s activities are attributed to an employer, and whether the intermediary has its own obligations, requires analysis of the law and actual relationship. This reference does not claim automatic employer liability for all supplier acts. Public Act 103-0804.

Begin with four facts: who defines the selection criteria, who operates the tool, who interprets its output and who makes the employment decision. Then identify the records available at each stage. A contract’s label for a supplier may leave those practical facts unanswered.

Map the service chain

Create one row per meaningful stage of the workflow. A candidate may move through sourcing, screening, assessment, shortlisting, interview scheduling and final selection. Different organizations may operate those stages, and the supplier may rely on another technology provider.

  • Service. What employment task is being performed?
  • Organization. Who is responsible for that stage?
  • System and feature. What technology is active, including subcontracted components?
  • Inputs and outputs. What information goes in and what recommendation, score or exclusion comes out?
  • Decision. How does a person use the output, and can it be overridden?
  • Record. Where is evidence retained and who can obtain it?

Do not infer AI use from a brand name alone. Ask about the service configuration actually used for your account. Record a supplier’s “no AI used” response with a date, the service covered and the person responsible for notifying you if that changes.

Questions for the agency and its technology supplier

Send a focused request tied to the relevant service. Ask the agency to identify automated functions that influence candidate screening, scoring, matching or ranking. Ask whether your criteria are supplied to the system and whether the agency changes or supplements them.

Request available evaluation information for the relevant use and population, material limitations, the handling of missing information and any location features. Ask how concerns are escalated and how a potentially affected decision can be reconstructed. If evidence sits with a subcontractor, record the route and expected response time for obtaining it.

Distinguish a response from supporting evidence. “We monitor fairness” is a statement about a process; it does not, without further information, show which measures were assessed, when, for which configuration or what action followed. Mark gaps explicitly instead of treating every returned questionnaire as a completed review.

Contract topics to review

The following are suggested review topics, not universal clauses expressly mandated by HB 3773. The agreement should reflect the actual service and the organization’s professional review.

  1. Disclosure of technology use. Define which functions and material changes the agency will communicate.
  2. Information access. Address the records the employer needs, available formats, response ownership and subcontractor dependencies.
  3. Communication responsibilities. Identify who prepares, reviews, delivers and records notices when they are relevant to the workflow.
  4. Concerns and changes. Establish escalation for incidents, disputed outcomes and significant system changes.
  5. Exit and continuity. Address record access, transition support and data handling if the service ends.
  6. Risk allocation. Have qualified reviewers assess the interaction of indemnities, limitations and applicable obligations.

A contractual allocation of tasks does not by itself determine regulatory responsibility. Conversely, the existence of a supplier relationship should not be treated as proof of a violation. Keep the contractual commitment and the actual operational practice side by side.

Coordinate communication without importing withdrawn rules

The statutory notice obligation remains relevant to covered employer use. IDHR’s proposed Part 2520 procedures were formally withdrawn in the June 26, 2026 Illinois Register. A proposed notice format or timing rule is not an adopted requirement. 50 Ill. Reg. 8755.

For an agency-mediated workflow, identify the person or team that can explain the use accurately. Decide how a change in screening criteria or technology reaches the notice owner. Record what communication was used and for which version of the process. Treat pre-use coordination as a practical design recommendation, while checking the current legal requirements for the actual workflow.

A fictional worked example

A fictional Illinois employer uses a staffing firm to shortlist warehouse supervisors. The firm says its platform ranks candidates, but the employer does not know whether the ranking uses location information or whether recruiters see excluded candidates.

The employer records those unknowns, requests the active feature description and asks the firm to explain how the ranking affects shortlisting. Procurement identifies the contract’s information-access clause. HR maps who handles candidate questions. The review owner records the evidence received, the remaining gaps and the decision about next steps.

The example demonstrates a documentation process. It does not establish that the fictional tool is lawful, unlawful, discriminatory or safe. A template should keep those conclusions open until the relevant facts and standards have been assessed.

A manageable review sequence

First, identify the agencies that affect a priority employment decision. Second, map the actual service chain and request evidence. Third, review unresolved items with the responsible HR, technical, procurement and professional contacts. Fourth, document changes, communication responsibilities and the next review trigger.

Repeat this work when the agency introduces a material feature, changes its technology supplier or changes how people are evaluated. Keep the latest decision record easy to locate, and preserve earlier versions according to the applicable retention and preservation process.

The TalentSight library includes vendor questions, a response tracker and related working documents. Use them to organize evidence and follow-up. A completed file is not a legal determination about the relationship.

How to cite this article

APA

Abdullahi, K. M. (2026, May 26). AI Used by Recruiting Agencies: An Illinois Employer Diligence Reference. Techné AI. https://techne.ai/insights/agency-liability-hb-3773

MLA

Abdullahi, Khullani M. "AI Used by Recruiting Agencies: An Illinois Employer Diligence Reference." Techné AI, May 26, 2026, https://techne.ai/insights/agency-liability-hb-3773.

Plain text

Abdullahi, Khullani M. "AI Used by Recruiting Agencies: An Illinois Employer Diligence Reference." Techné AI, May 26, 2026. Available at: https://techne.ai/insights/agency-liability-hb-3773

Get the next piece

The Techné Institute is the firm’s periodical on AI, work, and the law, written for boards, GCs, and the advisors who serve them. Read the published archive and subscribe for new essays.

About the author

Khullani M. Abdullahi, JD, is an AI governance and compliance consultant and the founder of Techné AI, an independent advisory firm based in Chicago. She submitted written testimony to the Illinois Senate Executive Subcommittee on AI and Social Media. She authored the AI Governance & D&O Liability briefing, maintains the Illinois AI Legislative Ecosystem tracker, and hosts the AI in Chicago podcast. Techné AI is an advisory firm, not a law firm.