What is an automated employment decision tool (AEDT)?
If a tool scores or ranks candidates for you, New York City law may already require a bias audit and advance notice before it runs. Most teams first meet the term inside a question they were expected to have answered already.
Automated employment decision tool (AEDT)
AEDT
An automated employment decision tool (AEDT) is a computational process derived from machine learning, statistical modeling, data analytics or artificial intelligence. It issues a simplified output: a score, classification or recommendation. That output is used to substantially assist or replace discretionary decisions about hiring or promotion. The term comes from New York City's Local Law 144, the first US law to require bias audits of such tools.
Read the definition as a three-part test. First, a computational process built on machine learning, statistics, data analytics or AI. Second, a simplified output: one score, one class, one recommendation. Third, that output substantially assists or replaces a person's call on hiring or promotion. Ranking engines, knockout scores and video-interview scoring all pass the test.
Plenty of hiring software fails it, which is the useful part. A parser that lifts fields from a resume without screening anyone issues no simplified output. A scheduling assistant decides nothing about the candidate. Regulation starts where the tool's number starts doing the recruiter's job. You'll also see the press write it as an algorithmic hiring tool; the term in the law is automated.
The duties once a tool qualifies
Screen for a New York City role with an AEDT and you owe a bias audit from the past year. A public summary of the results goes on your website. Candidates get notice at least 10 business days before the tool runs on them. The law covers employers and employment agencies alike. Enforcement by the city's Department of Consumer and Worker Protection began on 5 July 2023.
Penalties under the DCWP schedule reach 500 dollars for a first violation and up to 1,500 dollars for each one after, per day of use. And the concept travels. The EU AI Act puts hiring and worker-management AI in its high-risk class, and several US states now police automated decisions at work. Classify a tool for New York once, and you'll likely reuse the work elsewhere.
- Computational processML, statistics, data analytics or AI.
- Simplified outputA score, class or recommendation.
- Substantial assistDrives or replaces the human call.
- Covered useHiring or promotion, NYC role.
- Duties attachAudit, summary, 10-day notice.
Fail any step and the tool sits outside Local Law 144. Pass them all and the duties attach.
Related questions
01How is an AEDT different from an ATS?+
An applicant tracking system stores and moves candidate records; that alone triggers nothing. It crosses into AEDT territory when it scores, ranks or filters candidates with a model, and many modern ATS products bundle both. Classify the feature, not the product name.
02Does using AI anywhere in hiring make it an AEDT?+
No. The tool has to issue a simplified output that substantially assists or replaces the decision. Interview transcribers, schedulers and sourcing assistants sit outside the definition. They score no one, though other laws can still apply to them.
03Is the term used outside New York?+
New York City wrote the definition, but the concept travels. AEDT is the word US regulators and journalists reach for when they mean AI hiring tools. The EU AI Act puts the same tool class in its high-risk category, so expect the three-part test to shape other laws.
Some of the systems we have shipped
Related
- Does NYC Local Law 144 apply to a staffing agency's screening tools? →The compliance duties, mapped for agencies.
- Is AI resume screening safe to use? →Bias and accuracy, the question underneath the definition.
- How to resurface candidates in your ATS →Ranking with reasons, on the right side of the line.
- Software for staffing agencies →Our staffing hub: what agencies actually build.
