New York’s AI Labor Information Act Isn’t About Candidates. It’s About Whether You Can Explain Your Own Hiring.

Every AI-hiring law we’ve covered so far, NYC’s Local Law 144, Connecticut’s CART Act, Colorado’s SB 205, has been built around the same idea: tell the candidate they’re talking to a machine. New York’s newest bill skips that entirely. The AI Labor Information Act, sponsored by State Senator Michelle Hinchey and passed by the New York Legislature in June 2026, doesn’t ask employers to disclose anything to applicants. It asks them to report to the state, once a year, exactly how AI changed who got hired, laid off, or had their hours cut. As of late June 2026 it’s sitting on Governor Hochul’s desk, awaiting signature. If it lands, it becomes the first mandatory AI labor-impact reporting requirement in the country.

What the bill actually requires

Strip away the headlines and the obligation is specific. Covered employers, publicly traded companies and businesses with more than 50 employees based in New York, would have to file an annual report disclosing: an estimate of how many employees were displaced, hired, or had their hours increased or decreased because of AI use; a plain description of why the company deployed AI in the first place; what human oversight sits on top of the system; and how the AI touches sensitive personal data. None of that is candidate-facing. There’s no banner, no consent checkbox, no “you are speaking with an AI system” script. The report goes to the state, not the applicant.

That’s a meaningfully different compliance posture from every other AI-hiring law on the books right now, and it’s the part most GCC and India-facing compliance teams are going to miss, because we’ve spent the last year building disclosure checklists for NYC and Connecticut, and this bill doesn’t fit that template.

Why “we already have a disclosure banner” doesn’t cover this

A disclosure banner answers the question “does the candidate know an AI is involved.” This bill asks a different question entirely: “can the employer, as an organization, produce a number.” Not a policy statement. A number. How many hires this year were AI-influenced. How many roles saw hours cut because of it. That requires the kind of instrumentation most screening stacks were never built for, a running count of AI-touched decisions, tied to outcomes, aggregated annually, ready to hand to a regulator on request.

For a GCC or an Indian IT services firm sending campus hires into New York-based teams, or a company headquartered in New York running its screening through an offshore vendor, this is the harder ask. You can bolt a disclosure sentence onto an interview script in an afternoon. You cannot retroactively reconstruct a year’s worth of AI-influenced hiring decisions if nobody was counting them as they happened. We’ve run pilot campaigns where a single evening produced 3,000 candidate interviews; the operational discipline this bill demands is knowing, cleanly, which of those interviews fed into an actual hiring outcome and which didn’t, not scrambling to reconstruct it in March.

The part that should worry vendors more than employers

Most AI-hiring compliance conversations assume the employer carries the risk and the vendor just needs to stay out of the way. This bill’s reporting burden sits with the employer, but the data it requires, how many candidates the AI screened, how many it recommended for hire, how that split shifted headcount, lives entirely inside the vendor’s system. An employer can’t file an honest annual report if their screening vendor can’t hand over clean, auditable numbers on demand. If your AI-screening vendor’s answer to “how many of our hires this year were meaningfully shaped by your tool” is a shrug, that’s now a New York compliance gap wearing a vendor-selection problem’s clothes.

What to do before this gets signed

Don’t wait for the signature to start the inventory. Whether or not Hochul signs this specific bill, the direction is set, first NYC required disclosure, now New York State is testing outcome reporting, and other states will copy whichever version survives its first enforcement cycle. The employers in the best position aren’t the ones with the fastest disclosure banner. They’re the ones who can already answer, from their own system logs, how many candidates their AI screened last quarter and what happened to them. That’s not a legal question. It’s a data-hygiene question, and it’s cheaper to answer now than to reconstruct under deadline pressure next March.

Talk to us if you want a second read on whether your current screening vendor could hand you a clean, auditable answer to “how many hires were AI-influenced this year” if a regulator asked tomorrow.

Sources: Thompson Coburn — New York Legislature Passes AI Workforce Impact Disclosure Bill, Fisher Phillips — New York Employers Need to Prepare for 3 New AI Laws