Do these AI hiring laws apply to my firm even though we didn't build the tool?

Yes. That is the trap. Laws in New York, California, Illinois, and Connecticut name staffing agencies and employment agents as responsible parties. You use the tool to screen or rank candidates, so you carry the obligation. The vendor building it does not shield you. Read the actual statute language for each state you place in, because most of them use the word 'agency' on purpose.

How much does a bias audit actually cost?

For a single AI screening tool, a third-party audit usually runs between 5,000 and 25,000 dollars depending on the vendor and the volume of data. That sounds like a lot until you compare it to one discrimination claim. A single class action tied to biased screening can cost six or seven figures in legal fees alone. The audit is the cheap option.

What if my ATS vendor says they already handle compliance?

Get it in writing and read it carefully. Most vendor 'compliance' language covers the tool itself, not your use of it or your candidate disclosures. Under Local Law 144, the employer or agency using the tool is responsible for posting the disclosure and publishing the audit summary. Your vendor can hand you the audit, but you still have to act on it. Never assume a sales rep's verbal promise covers your legal exposure.

AI Recruiting Tools: The Bias Audit You Can't Skip

New AI hiring laws in New York, Colorado, California, Illinois, and Connecticut name staffing agencies as responsible parties, not just their clients. If you haven't audited the AI tools in your ATS, your firm carries the exposure. Here is what to fix this week.

Every AI screening tool in your ATS stack is now a legal question, not just a tech feature. If you place candidates in New York, Colorado, California, Illinois, or Connecticut, the law already names you as a responsible party. Not your client. Not your software vendor. You.

Most staffing VPs I talk to think this is an end-employer problem. They assume the company hiring the worker owns the compliance risk. That assumption is wrong, and it is getting more expensive by the month.

The Laws Changed and They Named You On Purpose

Here is what the last two years produced. NYC Local Law 144 requires a bias audit and public disclosure for any automated employment decision tool used to screen candidates in New York City. Colorado's AI Act is now in force and covers high-risk AI systems used in hiring. California, Illinois, and Connecticut passed their own rules.

The detail that matters: these laws do not stop at the employer. Several of them use the words 'employment agency' and 'employment agent' directly. That is staffing language. When your recruiter runs a candidate through an AI ranking tool, your firm is making an automated employment decision. The statute treats you as the party responsible for the audit and the disclosure.

You did not build the tool. Does not matter. You used it. The law follows the use, not the code.

So the personal exposure lands on the VP who approved the tech stack. If a candidate files a complaint about biased screening, the regulator does not call your software vendor first. They call the agency that ran the screen.

The One Thing You Have To Do

Get a bias audit on every AI tool in your stack that screens, ranks, or scores candidates. Then publish what the law requires you to publish.

That is the one thing. It sounds simple. Most firms have not done it because nobody owns it internally, and everyone assumes someone else already handled it.

A real bias audit is not a checkbox. It is an independent analysis of how your tool performs across race, gender, and other protected categories. It measures whether the tool passes qualified candidates from one group at a lower rate than another. If it does, you have a problem you need to document and fix before you keep using it.

Here is what a proper audit process looks like for a staffing firm:

Inventory every AI tool that touches a candidate. Sourcing tools, resume rankers, chatbot screeners, video interview scorers. If it makes or influences a decision, it counts.
Confirm which laws apply based on where you place people. Not where your office sits. Where the candidate works.
Get the audit from an independent third party. The vendor cannot audit their own tool and satisfy Local Law 144. It has to be independent.
Publish the required disclosures. Local Law 144 requires a public summary of the audit results and advance notice to candidates. Colorado has its own notice rules.
Give candidates the notice before you screen them. This is the step firms forget. The disclosure has to happen before the tool runs, not after.

Notice that most of this work is not technical. It is operational. Somebody on your team has to own the inventory, chase the vendors, and manage the disclosures. That is where firms fall down.

Why Your Vendor Will Not Save You

ATS and AI vendors love to say they are 'compliant.' Read what that actually means. Almost always, it means the tool itself has some documentation. It does not mean your use of the tool is covered.

The disclosure obligation is yours. The audit publication is yours. The candidate notice is yours. Your vendor can hand you a report, but they cannot post it on your careers page or send the notice to your candidates. You do that.

I have watched firms buy an expensive AI screening tool, trust the sales rep's promise that 'everything is handled,' and then discover during a client audit that no disclosure ever went out. The client's legal team found it. The staffing firm looked careless, and the contract was at risk.

PE-backed portfolios have an extra reason to care. Compliance gaps like this show up in due diligence. A buyer's counsel will ask whether your AI tools have been audited. If the answer is no, that is a mark against your valuation and a reason for a lower offer or an escrow holdback.

The exposure is not hypothetical. Enforcement in New York is picking up. Colorado's law is live. The trend across states is clear, and it points at agencies.

What This Costs You If You Ignore It

Run the math. A third-party bias audit for one tool costs somewhere between 5,000 and 25,000 dollars. A discrimination claim tied to biased screening costs far more. Legal defense alone can hit six figures before you get anywhere near a settlement.

Then add the client damage. Enterprise clients now put AI compliance language in their staffing agreements. If you cannot prove your tools are audited, you lose the account or never win it in the first place.

The audit is cheap insurance. The claim is the disaster. Firms that treat this as a cost center are looking at it backward.

Do This Week

Pull a list of every AI tool in your stack that screens, ranks, or scores candidates. Put one name next to each tool as the owner. Then send one email to each vendor asking two questions: Has this tool completed an independent bias audit, and can you send me the results and the disclosure language I need to publish.

You will learn fast which vendors have their act together and which ones are bluffing. That list alone tells you where your exposure sits. Start there, and you turn a legal liability into a controlled process before a regulator or a client does it for you.

Your AI Recruiting Tool Is a Legal Liability If You Haven't Done This One Thing