Illinois HB 3773 has been live since January 2026. If you run AI screening in Illinois with no governance policy, you are already operating out of compliance. That is not a warning about the future. That is today.
Colorado SB 26-189 lands in January 2027. More states are drafting their own versions right now. Every one of them has slightly different rules, different definitions, and different penalties. You are not looking at one law to follow. You are looking at a map that keeps adding borders.
Here is the part that should get your attention. Most staffing firms have no plan for this. Only about a quarter of staffing organizations have a documented AI governance policy. The rest are deploying AI screening at scale and hoping nobody asks how it works.
When the question comes, and it will, hope is not an answer.
Why staffing firms became the target
You sit in the middle of every hiring decision. That is the whole business. A client needs people, you find and screen them, you place them. AI speeds that up, so firms turned it on across sourcing, screening, and ranking.
That scale is exactly what creates the exposure. A single AI platform running across your operation can reach thousands of candidates a week. The American Staffing Association reports that when an algorithm produces a discriminatory effect, three parties may share the exposure: the technology provider, the staffing firm, and the client.
Read that again. The staffing firm is in the middle of the liability, not off to the side. You cannot point at the tool and walk away. You chose it, you ran it, and you handed the output to your client.
So when a law like HB 3773 says AI cannot discriminate in hiring, it is talking about the screening you already run. Not a hypothetical. The req you filled yesterday.
The laws cover more of your workflow than you think
Most leaders assume these laws only cover the final hire decision. They do not. The scope is much wider.
Staffing Industry Analysts reports that AI tools used anywhere in the talent lifecycle must be assessed through the same compliance lens you have always applied to human decision-makers. That includes sourcing, screening, ranking, assignment, performance evaluation, and termination.
Walk through what that means for your operation.
- An AI tool that ranks candidates for a req? Covered.
- A sourcing tool that decides which profiles surface first? Covered.
- An AI that scores contractor performance mid-assignment? Covered.
- Automation that flags someone for the end of an assignment? Covered.
If a human doing that job would have to follow anti-discrimination rules, the AI doing it has to as well. The law does not care that it is software. It cares about the outcome.
This is where most firms get caught. They turned on a ranking feature two years ago to save recruiters time. Nobody documented why. Nobody checked who it favored or filtered out. It just runs. Under these laws, that quiet tool is a live risk sitting in your workflow.
What the compliance gap actually costs you
The 25% number is the real story. Three out of four staffing organizations are running AI without a written governance policy. That gap is not just legal risk. It is operational risk, and it hits in three ways.
First, you cannot answer basic questions. A regulator or a client asks which AI tools you use and how you check them for bias. If you cannot produce a document, you look negligent even if your tools are fine.
Second, your clients are starting to ask first. PE-backed buyers and large enterprise clients now put AI questions in their vendor reviews. No policy means you fail the review and lose the deal. The compliance gap costs you revenue before it ever costs you a lawsuit.
Third, you carry shared liability you never priced in. Remember, the exposure splits three ways. If your client gets sued over a discriminatory screen, their lawyers will look straight at the firm that ran it. You want a paper trail that shows you did the work, not silence that makes you the easiest target.
What to do about it this week
You do not need a legal department to start. You need a record. Here is the move.
Build your AI inventory. Open a spreadsheet. List every tool in your operation that touches a candidate or contractor decision. Sourcing, screening, ranking, assignment, performance, termination. For each one, write down three things: what it does, who approved it, and whether it makes or influences a decision about a person.
That list is the foundation of everything. You cannot govern what you have not named. Most firms discover two or three tools they forgot were even running.
Then do the hard part. For each tool that influences a decision, ask the provider for their bias testing and documentation. If they cannot give it to you, that is a flag you need to know about now, not after a complaint lands.
This is not a one-time project. Illinois is live, Colorado is coming, and more states are drafting. The patchwork grows every quarter. The firms that win are the ones who treat AI governance as a standing operation, not a fire drill.
Your action this week: Build the AI inventory. One spreadsheet, every tool, who approved it. If you do nothing else, do that. It is the first document a regulator asks for and the first thing a client wants to see. You can have it done by Friday.
If you want help turning that inventory into a governance policy your clients and regulators will accept, that is the work we do at Leap. Book a discovery call and we will map your exposure together.