Illinois HB 3773 took effect January 1, 2026. Connecticut's CART Act provisions began October 1, 2026. These are not drafts or proposals. They are live law, and your firm is already operating under them.
Most staffing leaders I talk to still file AI compliance under "we'll deal with it later." That is the mistake. The clock already started. Every candidate your AI screening tool touched this quarter is a record you may have to defend, and the laws were written in a way that puts staffing firms on the hook more than the clients they serve.
Let me walk you through what changed, why it lands harder on you than on your clients, and the specific steps to take this week.
Why these laws point straight at staffing firms
Illinois amended its Human Rights Act in 2026 to prohibit employer use of AI that produces a discriminatory effect on applicants and employees. The amendment also adds candidate notification and record retention requirements. There is no federal framework to override it, according to the American Staffing Association. That last part matters. You cannot wait for Washington to settle this. State law is the law that applies to you today.
Here is the part that should get your attention. The Illinois law applies to employers and their agents, including recruiters and other third parties acting on an employer's behalf. That means staffing firms using AI are more at risk than the employers they serve, as StaffingHub explains. You run the screening. You touch the candidate. You are the agent. When a complaint comes, your name is on the process.
Connecticut's CART Act adds another layer in a neighboring market. If you place candidates across state lines, you do not get to pick one rulebook. You have to meet the toughest standard that applies to any candidate you handle.
And this is spreading. The same StaffingHub reporting notes NYC mandates independent bias audits with fines up to $1,500 a day. That is a daily meter running against a firm that gets it wrong. Three jurisdictions, three sets of rules, one firm trying to scale on AI. That is the reality now.
The scaling problem hiding inside the compliance problem
Here is why this belongs in a conversation about scaling, not just legal. AI is how modern staffing firms grow without adding headcount at the same rate as revenue. You screen more candidates, rank faster, and move reqs quicker. That leverage is real.
But the same AI that scales your desk also scales your exposure. One screening rule that quietly filters out a protected group does not hurt one candidate. It hurts thousands, across every req it touched, in every state where you placed. The tool that made you faster made the problem bigger too.
So the firms that treat compliance as a bolt-on at the end are building on cracked ground. The firms that build it into how the AI runs are the ones who can actually scale without a lawsuit waiting at the top.
This is a data problem as much as a legal one. If you cannot show how your AI reached a decision, you cannot defend it. If your candidate records live in three systems that do not talk to each other, you cannot produce a clean record when a regulator asks. Clean data and clear documentation are not paperwork. They are the thing standing between you and a daily fine.
What to do this week
You do not need a six-month project. You need to start, and you need to start now. Here is the order I would run it.
1. Build the inventory. List every place AI touches a candidate in your process. Sourcing, resume ranking, screening questions, scheduling, scoring. If a tool filters, ranks, or scores a person, it goes on the list. You cannot comply with rules for tools you have not named.
2. Flag the high-risk uses. Anything that decides who moves forward is high risk. That is where a discriminatory effect shows up. Put a star next to every one of those.
3. Turn on candidate notification. Illinois requires it. Write a plain notice telling candidates that AI is part of your process. Put it in your application flow. Put it in your recruiter scripts. Make it automatic so it happens every time, not when someone remembers.
4. Fix your record retention. Decide where AI decision records live and how long you keep them. One system, one standard. If a complaint comes, you want to pull the record in an hour, not spend a week hunting across platforms.
5. Confirm your screening tools have been checked for bias. Ask your tool providers directly for their bias testing and audit documentation. If you place into NYC, independent audits are not optional. Get those on file.
None of this requires you to stop using AI. That is the point I want to land. The goal is not to slow down. It is to scale on AI you can actually stand behind when someone asks hard questions.
The cost of waiting
The firms that lose here are not the ones using AI. They are the ones using it blind. No inventory, no notice, no records, no idea which tool scored which candidate or why. When a complaint hits that firm, they have nothing to show. They are the agent named in the law with no defense built.
You scale faster than a client can. You touch more candidates. You carry the exposure. That is the trade that came with the leverage, and ignoring it does not make it go away. It just makes the bill bigger when it arrives.
This week, pull your leadership team together for one hour and build the inventory. Just the list. Every place AI touches a candidate in your process. That single document is the foundation for everything else, and most firms do not have it yet. Start there, and you are already ahead of the pack still treating this as a future problem.