Automated Employment Decision Tool (AEDT)

An automated employment decision tool, or AEDT, is software that uses machine learning, statistical modeling, data analytics or artificial intelligence to produce a score, classification or recommendation that substantially assists or replaces a human decision about hiring or promotion.

The term is a legal one. It was defined by New York City's Local Law 144 of 2021, the first law in the United States to regulate the use of such tools directly, and it has since been borrowed by other jurisdictions. Résumé screeners that rank applicants, video interview scorers, assessment platforms and any system that produces a shortlist are the typical examples.

What Local Law 144 requires

The law took effect on 1 January 2023 and the Department of Consumer and Worker Protection began enforcing it on 5 July 2023. An employer or employment agency may not use an AEDT to screen candidates or employees for a position in New York City unless three conditions are met.

1. A bias audit has been carried out by an independent auditor within the year before the tool is used. The audit calculates selection rates and impact ratios by sex, by race and ethnicity, and by the two combined, using the tool's own historical data or test data where that is insufficient.
2. A summary of the audit is published on the employer's website, along with the distribution date of the tool, and stays up for as long as the tool is in use and six months afterwards.
3. Notice is given to candidates and employees at least ten business days before the tool is used, stating that an AEDT will be used, which job qualifications and characteristics it will assess, and how to request an alternative process or accommodation. The notice also has to explain what data is collected, its source and the retention policy, or say where that information can be requested.

The word doing the most work in the definition is "substantially". A tool whose output is one input among several that a human weighs is arguably outside the law; a tool whose score decides who a human ever sees is inside it. The DCWP rules resolve this by treating a tool as covered when its output is relied on to overrule or outweigh other factors, or is the only factor used.

How enforcement has gone

Slowly, and then critically. In December 2025 the New York State Comptroller published an audit of DCWP's enforcement of the law. It found that of 32 bias audits DCWP had reviewed, the agency identified one compliance issue, whereas the Comptroller's own review of the same 32 found 17 potential violations. Three quarters of test calls to the city's 311 line about AEDT complaints were misrouted and never reached DCWP. The report described the enforcement system as ineffective, and DCWP committed to changes to intake, training and investigation.

The practical reading for an employer is not that the law can be ignored. It is that the published audit summaries have been reviewed lightly so far and will be reviewed more carefully from here, and that a summary which would not survive a second look is now a liability rather than a formality.

Beyond New York City

The pattern is spreading, with different definitions and different mechanisms.

• Illinois amended its Human Rights Act, effective 1 January 2026, to make it a civil rights violation for an employer to use artificial intelligence in a way that has a discriminatory effect on a protected class, and to require notice to employees when AI is used in employment decisions.
• Colorado enacted the first state-wide law on high-risk AI systems, and California has its own set of workplace AI bills; including those used in employment decisions, with duties of care on both developers and deployers. Its effective date has been pushed back once; check the current status before relying on any date.
• Federal guidance from the EEOC on AI in selection procedures was withdrawn in early 2025. Title VII, the ADA and the Age Discrimination in Employment Act continue to apply to the outcomes of any tool regardless, which is where the disparate impact analysis in a bias audit comes from in the first place.

An employer using the same screening tool in several states is, in practice, best served by meeting the New York City standard everywhere: it is the most specific, and an audit that satisfies it goes most of the way to satisfying the others.

Where the definition stops, and where the same questions continue

Local Law 144 covers decisions about hiring and promotion. It does not cover tools that score current employees' productivity, flag them for performance management, or rank them for redundancy, even though those tools do the same thing with the same risks. That gap is where most of the growth in workplace AI is happening, and it is the subject of our guide to agentic AI in HR.

The questions a bias audit asks are the right ones for those tools too: what the tool measures, whether the outcome differs by protected group, whether a human can see and override it, and whether the people it scores were told. Our performance tracking guide and the state monitoring law summary cover the parts of that which are already regulated.

Key points

• An AEDT is software using machine learning, statistics or AI to produce a score or recommendation that substantially assists or replaces a hiring or promotion decision.
• NYC Local Law 144 requires an independent bias audit within the past year, a published summary, and notice to candidates ten business days before use.
• The bias audit calculates selection rates and impact ratios by sex, by race and ethnicity, and by the intersection of both.
• The December 2025 State Comptroller audit found DCWP identified one issue in 32 audits where the Comptroller found 17, and called enforcement ineffective. Expect closer review from here.
• Illinois made discriminatory AI use a civil rights violation from 1 January 2026, with a notice requirement. Colorado's law covers high-risk AI in employment.
• The law stops at hiring and promotion. Tools that score current employees carry the same risks and are covered by the same anti-discrimination statutes.