AI Hiring Regulation Is Here: What NYC Local Law 144 and the EU AI Act Mean for Your Funnel

For two years, AI hiring regulation was something to watch. In 2026, it is something to comply with. Two regimes matter most — and both put the burden on the employer, not the software vendor.

NYC Local Law 144: enforcement is waking up

New York City's Local Law 144 requires any employer using an automated employment decision tool (AEDT) to screen NYC candidates to:

• Commission an annual independent bias audit and publish a summary of the results publicly.
• Give candidates at least 10 business days' notice before an AEDT is used on them, including what qualifications and characteristics it evaluates.
• Accept that liability sits with the employer, not the vendor — buying a tool does not transfer the obligation.

Penalties run $500 for a first violation and up to $1,500 for each subsequent one — assessed per day, per affected candidate, which compounds quickly at hiring-funnel volume. Enforcement to date has been light, but a recent NYC Comptroller audit called the current enforcement regime "ineffective" and the enforcement agency has committed to tightening — meaning the grace period is ending.

The EU AI Act: August 2, 2026

The EU AI Act classifies AI systems used in recruitment, screening, hiring, promotion, and termination as high-risk, with the full high-risk obligations becoming enforceable on August 2, 2026. Requirements include conformity assessment, human oversight, technical documentation available to regulators, and continuous post-market monitoring.

Two details US companies miss: the Act applies whenever the AI system's outputs are used in the EU — a US company screening candidates for a Dublin office is in scope — and fines for the most serious violations reach €35 million or 7% of global turnover.

What compliance actually requires of your funnel

Strip away the legal language and both regimes converge on the same demands:

1. Inventory. Know every place AI touches a hiring decision — including features buried inside your ATS that legal has never seen.
2. Transparency. Candidates must be told, in advance and in plain language, when AI evaluates them and on what criteria.
3. Auditability. You must be able to show how a candidate was scored — which means structured criteria and retained records, not a black-box ranking.
4. Human oversight. A human must be able to review and override AI-driven outcomes, and the review must be real, not a rubber stamp.
5. Bias monitoring. Regular analysis of outcomes across demographic groups, independently audited where LL144 applies.

The strategic read

It is tempting to read this as a reason to avoid AI in hiring. The data points the other way: the requirements — documented criteria, consistent process, reviewable records, monitored outcomes — are easier to satisfy with structured, rubric-scored AI rounds than with the unstructured human interviews they replace. Nobody can produce an audit trail for what happened in a hallway debrief; a recorded, transcribed, rubric-anchored interview is its own audit trail.

The teams at risk are not the ones using AI deliberately. They are the ones using it accidentally — through unvetted ATS features and resume rankers — with no inventory, no notice, and no records. Regulation is not the end of AI hiring. It is the end of unstructured hiring wearing an AI costume.

This article is general information, not legal advice — loop in employment counsel for your specific obligations.