US State AI Hiring Laws: What Applies Where

The United States has no single AI hiring law. It has a patchwork: New York City requires bias audits and candidate notice for automated employment decision tools, Illinois regulates AI-analysed video interviews, and Colorado has a broader AI Act. The obligations follow where the role is, which means an agency serving US clients from India needs to know which rules attach to which requisition.

Last reviewed August 2026. Summary of published statutes. Take legal advice before acting.

New York City Local Law 144

NYC Local Law 144 regulates automated employment decision tools used to screen candidates for positions in New York City.

Two headline requirements. A bias audit must be conducted, by an independent auditor, and a summary of results published. And candidates must be notified that an automated tool is being used, in advance, with information about what it assesses.

For a staffing firm the practical question is whether the tool you use has been audited and whether that audit is available to you. If your vendor cannot produce one, you cannot use the tool on NYC roles without carrying the exposure yourself.

Illinois and video interviews

Illinois AIVIA addresses AI-analysed video interviews specifically, which makes it the most directly relevant statute for firms running AI video screening.

The requirements centre on consent and explanation: telling candidates that AI will analyse the interview, explaining how it works and what characteristics it evaluates, obtaining consent before proceeding, and honouring deletion requests within a defined period.

The explanation requirement is the one firms underestimate. Saying that AI is used is not the same as explaining what it assesses, and a vendor who cannot describe their scoring in plain language makes this obligation impossible to meet.

Colorado and the broader direction

The Colorado AI Act takes a wider approach, covering high-risk AI systems including those making consequential decisions in employment, with duties around reasonable care to avoid algorithmic discrimination.

It is structurally closer to the EU AI Act than to the narrower NYC and Illinois rules, which suggests where US regulation is heading: comprehensive frameworks rather than tool-specific statutes.

For planning purposes, a firm that builds its process to the stricter standard covers the looser ones. Building to each statute separately produces a compliance matrix nobody maintains.

How to handle this per requisition

The rules follow the role's location, which means the question is operational rather than strategic.

Three things worth capturing on every US requisition. Where the role is based, since that determines which rules apply. Whether AI screening will be used on it. And whether the client has their own policy, which is frequently stricter than the statute and arrives as a contractual term rather than a legal one.

Firms that capture location and AI-use as structured fields on the requisition can answer this instantly. Firms that do not are checking manually on every role, which means they are not checking.

What to ask your vendor

Five questions, and a vendor unable to answer them is a vendor you cannot use on regulated roles.

Has the tool been bias audited, by whom, and how recently? Independence matters for NYC.

Is the audit summary available to customers? You may need to publish or reference it.

Can you explain in plain language what the system assesses? Required by Illinois, useful everywhere.

What is retained, for how long, and can it be deleted on request?

Where is the human checkpoint? The common thread across every framework, and covered in what cannot be delegated.

Frequently asked questions

The two fields to add to your requisition form

Role location and whether AI screening will be used. Both as structured fields, on every US requisition, captured at intake.

With those two, the compliance question answers itself on every role. Without them, somebody is supposed to be checking manually, which means nobody is.