New York Bill Would Force Employers to Report AI's Job Impact

New York lawmakers have passed legislation that would require most businesses operating in the state to report annually on how artificial intelligence affects their workforce—a pioneering attempt to quantify AI's employment impact that could reveal just how difficult such measurement actually is.

Assembly Bill A9581B, which passed both legislative chambers in June and awaits Governor Kathy Hochul's signature, would apply to businesses with more than 50 employees doing business in New York, as well as publicly traded companies. Covered employers would submit annual reports to the Department of Labor estimating how many workers were displaced or saw reduced hours because of AI, how many were hired or gained hours, and how many previously filled positions went unfilled due to the technology.

Why it matters

While employers increasingly cite AI in layoff announcements—more than 100,000 job cuts through June 2026, according to Challenger, Gray & Christmas—no state has systematically tracked AI's broader workforce effects beyond mass layoffs. New York's approach could create the first recurring dataset showing how AI changes employment through hiring decisions, attrition, and restructuring, not just terminations. The resulting data could inform workforce policy, retraining programs, and future AI regulation. But the legislation's usefulness hinges on a deceptively complex question: how do employers determine which workforce changes AI actually caused?

Building on WARN notice disclosures

New York has already begun collecting limited AI employment data. In her 2025 State of the State agenda, Hochul directed the Department of Labor to require businesses filing Worker Adjustment and Retraining Notification (WARN) notices for mass layoffs to disclose whether AI played a role. The state's 2026 WARN data currently include one employment action expressly identifying artificial intelligence: Nespresso's notice affecting 46 workers lists both "Relocation of Business" and "Artificial Intelligence" as reasons, though the public record doesn't clarify how much each factor contributed.

A9581B would extend disclosure requirements far beyond WARN-covered layoffs, which capture only plant closings, mass layoffs, and similar large-scale actions requiring 90 days' notice.

Capturing invisible workforce changes

The legislation targets workforce shifts that never trigger layoff notices. When an employee leaves voluntarily and AI allows remaining staff to absorb that work, no layoff occurs—yet a position disappears. Companies might reduce hiring, reallocate duties, or create new roles to manage AI systems without conducting mass terminations.

The bill's sponsor memorandum acknowledges uncertainty about whether AI-related displacement occurs through direct layoffs or reduced hiring as workers depart naturally. By requiring annual reports on displacement, hiring changes, hours adjustments, and unfilled positions, A9581B attempts to capture these less visible effects.

Businesses would also report how they use AI, including objectives, human oversight, frequency, and risk mitigation measures. The Department of Labor would aggregate submissions and publish annual analyses by sector, geography, and business size. Penalties for non-compliance could reach $500 per day.

The causation problem

Determining when AI "causes" workforce changes introduces significant complexity. The bill asks for estimates of changes occurring "due in full or in part" to AI use—language that acknowledges multiple factors often drive employment decisions.

Consider a company that deploys generative AI while simultaneously cutting costs amid weak economic conditions. Over a year, 100 employees leave and only 70 positions are filled because managers believe AI-assisted staff can handle some work. How many unfilled positions resulted from AI versus financial pressures or routine restructuring?

Two companies experiencing identical workforce changes might reach different conclusions about AI's role. One might attribute an unfilled position to automation; another might characterize the same decision as ordinary attrition. The Department of Labor's eventual methodology—including standardized forms and definitions—could determine whether the statewide data provide meaningful insights or simply reflect inconsistent employer interpretations.

Challenger, Gray & Christmas has cautioned that its own figures on AI-related job cuts reflect employer announcements rather than precise causation measurements. New York's proposal attempts deeper analysis by capturing multiple workforce dimensions directly from employers, but those employers must still make underlying causal judgments.

If enacted and implemented effectively, the legislation could offer policymakers unprecedented visibility into whether AI-related displacement clusters in specific industries, whether companies create jobs alongside eliminations, and whether larger employment effects occur through layoffs or quietly unfilled positions. The data's usefulness will depend on whether thousands of businesses can consistently answer a deceptively simple question: when did AI actually cause a job to change?

These details were first reported by Alonzo Martinez in Forbes.