Automated Employment Decision Tool (AEDT)

An Automated Employment Decision Tool, or AEDT, is any computational process — including AI, machine learning, or statistical modeling — used to substantially assist or replace human decision-making in hiring, promotion, or termination. The term was popularized by NYC Local Law 144, the first US law to require public bias audits before employers can use one.

What Counts as an AEDT?

Not every piece of HR software qualifies. A tool becomes an AEDT when it generates a score, classification, or recommendation that meaningfully influences an employment outcome — resume screeners that rank candidates, video-interview tools that score responses, and skills-assessment platforms that filter applicants all fit the definition.

Tools that simply organize data without generating a decision-affecting output, such as a basic applicant tracking system used only for storage, typically fall outside the definition. The line often comes down to how much the output actually shapes who advances.

Which Laws Regulate AEDTs?

NYC Local Law 144 was first, but it's no longer alone. Illinois, Colorado, and Texas have all passed employment-AI statutes that either explicitly define AEDT-style tools or regulate the same category of software under different terminology. The EU AI Act classifies employment-related AI as "high-risk" under a similar logic. Regulators across jurisdictions are converging on the same core requirement: if software materially shapes a hiring or personnel decision, someone has to be able to explain and audit it.

Why Does AEDT Regulation Matter for HR?

Unaudited hiring algorithms can quietly encode bias from historical data — screening out candidates on characteristics correlated with, but not explicitly, protected traits. Regulators treat this as a variant of disparate impact, the same theory long applied to non-AI selection procedures. An AEDT that hasn't been tested for skewed outcomes is a liability regardless of intent.

How Should Employers Manage AEDT Compliance?

Three practices cover most jurisdictions at once: commission an independent bias audit before deployment, publish audit results where required, and give candidates notice — and often an alternative process — before an AEDT screens them. Vendors should be contractually required to supply the data needed to audit their own tools, since employers, not vendors, usually bear legal liability.

Compliance also means documentation discipline. AI HR compliance software can track audit dates, notice deadlines, and jurisdiction-specific requirements automatically, which matters once an employer operates in more than one AEDT-regulated state.

How Is an AEDT Different From General HR AI?

General HR AI — chatbots answering benefits questions, AI that drafts job descriptions — doesn't trigger AEDT rules because it doesn't determine who gets hired or promoted. The regulatory trigger is decisional influence, not AI use itself. A tool that recommends interview questions is not an AEDT; a tool that ranks candidates by "fit score" almost certainly is.

Discover how our HR solutions streamline onboarding, boost employee engagement, and simplify HR management

Request a Demo

Frequently Asked Questions

Q: Does an AEDT have to use AI or machine learning?
A: No. NYC's definition also covers simpler statistical models and rule-based scoring systems, as long as they substantially assist a hiring, promotion, or termination decision.

Q: Who is liable if an AEDT vendor's audit was flawed?
A: Typically the employer using the tool, not the vendor — which is why contracts should require vendors to share the data needed for an independent audit.

Q: Does using an AEDT in one state trigger rules in another?
A: Generally no — most AEDT laws apply based on where the job or candidate is located, so multi-state employers need jurisdiction-by-jurisdiction tracking.

Q: Can a candidate opt out of an AEDT screening?
A: Under several current laws, yes — employers must offer an alternative selection process or accommodation on request.

Q: How often must an AEDT be re-audited?
A: NYC requires an audit within one year before use; most emerging state laws expect re-audits whenever the tool or its training data materially changes.

Q: Is a resume keyword filter an AEDT?
A: If it meaningfully ranks or eliminates candidates rather than just organizing applications, yes — simple sorting without a decisional score usually is not.