An AEDT is the legal name for the AI tool you might already be using to sort a stack of résumés. If you’ve ever asked ChatGPT to rank candidates, or bought software that scores applicants, you’ve brushed up against a term that now carries real obligations in 2026 — and most small employers have never heard of it. Here’s what an automated employment decision tool actually is, when your AI use becomes one, and what the law expects when it does.
TL;DR. An AEDT (Automated Employment Decision Tool) is any AI or algorithmic system that substantially assists or replaces a discretionary hiring or promotion decision — screening, scoring, or ranking candidates. Using one can trigger legal duties: a bias audit, a public results summary, and advance notice to candidates (10 business days under NYC’s Local Law 144). Using ChatGPT to rank applicants can count as an AEDT; using it to draft a job post usually doesn’t.
Last reviewed: 2026-07-21
What is an AEDT?
An AEDT — automated employment decision tool — is software that uses machine learning, statistical modeling, or other AI to make, or heavily influence, decisions about who gets hired, promoted, or advanced. The term comes from New York City’s Local Law 144 of 2021, the first US law to regulate these tools directly, and it has since become the standard shorthand across a wave of 2026 employment-AI laws.
The definition turns on a specific phrase: a tool is an AEDT when it “substantially assists or replaces discretionary decision-making.” That’s the line that matters. A résumé-scoring system that ranks applicants and tells you who to interview is squarely an AEDT. A spell-checker is not. And a general tool like ChatGPT sits in between — it depends entirely on what you use it for, which is exactly where most small employers get caught off guard.
When does an AI tool become an AEDT?
Not every use of AI in hiring is regulated — only the uses that meaningfully drive the decision. The test is whether the tool is doing the deciding (or heavily shaping it) versus merely helping you work faster. Drafting is low-risk; screening, scoring, and ranking is where AEDT duties attach.
The reason the line sits there is fairness. When AI merely helps you write faster, it can’t discriminate against an applicant — there’s no applicant in the loop yet. The moment AI starts sorting people, a biased or opaque system can quietly filter out protected groups, which is the harm every one of these laws is built to prevent.
Which laws regulate AEDTs?
What began as one NYC rule is now a patchwork. Four regimes matter most in 2026, and a small employer hiring across state lines can touch several at once.
| Law | Effective | Core AEDT duty |
|---|---|---|
| NYC Local Law 144 | Enforced July 2023 | Independent bias audit within the past year, public summary of results, and candidate notice ≥10 business days ahead, with an alternative process on request |
| Illinois HB 3773 | January 1, 2026 | Notify applicants when AI is used in an employment decision; ban on AI that discriminates by protected class or ZIP code |
| Colorado AI Act (SB 24-205) | In flux — start pushed to mid-2026, enforcement stayed, replacement (SB 26-189) due Jan 1, 2027 | Impact assessments, disclosure, and a right to human review — once it actually takes effect |
| EU AI Act | Core duties August 2, 2026 | Recruitment AI is “high-risk”: risk management, human oversight, and disclosure to candidates — and it applies to US firms hiring EU-based candidates |
NYC’s Local Law 144 remains the template: it’s the one with a concrete, enforced procedure — audit, post, notify. Illinois added the notice duty for a whole state in 2026. Colorado’s more sweeping version is stuck in legislative limbo and shouldn’t be over-built for yet. And the EU AI Act reaches across the Atlantic, so even a US small business recruiting an EU candidate can fall under it.
Does using ChatGPT count as an AEDT?
This is the question that catches small employers, and the honest answer is: it depends on what you ask it to do. None of these laws ban ChatGPT in hiring — they regulate the decision-making use. If you paste in a stack of résumés and ask ChatGPT to score, rank, or filter candidates, and you act on that ranking, you are plausibly using it as an AEDT and the disclosure and oversight duties can apply. If you use it to write a job description, suggest interview questions, or summarize a single résumé for your own reading, you’re generally in the safe, unregulated zone.
The practical safeguard is the same one the laws are pushing toward anyway: keep AI in the assist seat and a human in the decide seat. Use AI to summarize and surface candidates for your review — never to auto-reject. The moment a model is the thing that ends someone’s candidacy, you’ve stepped into AEDT territory with all its obligations. Our guide on what you have to tell applicants when you screen with AI walks through the exact disclosure line and a human-in-the-loop screening prompt.
What this means for small business owners
If you’re a solo owner or an “HR-of-one” using ChatGPT casually, you probably don’t need to stop — you need to add three habits. Put one plain disclosure sentence on your job posts (“We use AI-assisted tools to help review applications; a human reviews every candidate”). Never let AI send rejections on its own. And keep a short record of what you used it for. That covers the heart of every AEDT rule without a compliance department. If you hire in NYC specifically and your tool “substantially” drives screening, you also need a bias audit — which is a reason to keep AI in the assist seat, where the audit trigger doesn’t bite. The ChatGPT for Business course covers where AI saves real time in operations and where it needs a human hand.
What this means for HR and recruiters
For anyone whose job is hiring, AEDT literacy is now core professional knowledge. The compliance pattern is consistent across the laws: audit the tool, disclose its use, keep human oversight, and offer an alternative. If you buy a dedicated screening or ranking platform, ask the vendor for its most recent independent bias-audit results before you deploy — under NYC law, the obligation to have that audit falls on you, the employer, not the vendor. And remember the demographic risk documented for AI in hiring: opaque scoring tends to disadvantage the same groups anti-discrimination law protects, which is both a legal and an ethical liability.
What this means for hiring managers
You may be using an AEDT without buying one — if your applicant-tracking system quietly ranks candidates, or if a teammate runs applicants through an AI scorer, that’s an AEDT in your process. Ask what’s actually happening to applications before they reach you, whether a required notice went out, and whether a human reviews every screen-out. A ranked shortlist feels neutral; the ranking underneath it may not be.
What this means for job applicants
If you’re applying for jobs, an AEDT is why you sometimes get a fast, impersonal rejection. In covered places you have rights: under NYC’s law, employers must notify you before using an AEDT and offer an alternative process on request. Knowing the term helps you ask the right question — “was an automated tool used to evaluate my application, and can I request a human review?” On the flip side, if you’re worried your own AI-assisted résumé might be misjudged, that’s a different tool (an AI detector) with its own well-documented reliability problems.
Common misconceptions
“AEDT only applies to big companies with fancy hiring software.” The definition is about the function — substantially assisting a hiring decision — not the price tag. A small business using a free AI tool to rank candidates can meet the definition just as a Fortune 500 platform does.
“If I use ChatGPT, I’m automatically breaking the law.” No. These laws regulate AI that drives hiring decisions, with disclosure and oversight. Assisted screening with a human deciding, plus a disclosure line, is permitted everywhere. Drafting a job post isn’t even in scope.
“The vendor handles compliance for me.” Under NYC’s Local Law 144, the duty to have a bias audit and give notice falls on the employer using the tool, not the software vendor. Buying a “compliant” tool doesn’t transfer your notice obligation.
“Colorado’s sweeping AI hiring law is in force now.” It isn’t. SB 24-205’s start date slipped to mid-2026, enforcement was stayed, and a replacement framework is scheduled for 2027. Don’t rebuild your whole process around a stayed law.
Frequently asked questions
What does AEDT stand for?
AEDT stands for Automated Employment Decision Tool. It’s any computational system — using AI, machine learning, or statistical modeling — that substantially assists or replaces a discretionary decision about hiring, promotion, or advancement. The term originates in New York City’s Local Law 144.
Does ChatGPT count as an AEDT?
It can, depending on use. If you use ChatGPT to score, rank, or filter job candidates and act on that output, it may qualify as an AEDT and trigger disclosure and oversight duties. If you use it to draft a job post, write interview questions, or summarize a single résumé for your own reading, it generally does not.
What do the AEDT laws actually require?
The common requirements are transparency and human oversight: disclose to candidates that AI is used (Illinois requires notice; NYC requires notice at least 10 business days ahead), keep a human genuinely in charge of decisions, and — under NYC’s Local Law 144 — complete an independent bias audit and post a public summary. The EU AI Act adds risk-management and documentation duties for high-risk hiring AI.
Do AEDT laws apply to small businesses?
Yes, where the tool meets the definition. NYC’s Local Law 144 has no small-employer exemption if you use an AEDT. Illinois’s notice duty applies to covered employers (generally 15+ employees) regardless of the tool’s cost. The EU AI Act offers some proportionality for small firms but does not exempt them from core high-risk duties.
Is using AI to screen résumés legal?
Generally yes, if you use it responsibly: disclose it, keep a human making the final call rather than auto-rejecting, avoid discriminatory outcomes, and — in covered jurisdictions — meet the audit and notice requirements. No current law bans AI-assisted hiring outright.
The bottom line
An AEDT is simply AI that helps decide who gets hired — and once you’re using AI that way, a growing set of 2026 laws asks for something reasonable: tell candidates, keep a human in charge, and (in NYC) audit the tool for bias. The safest posture for a small employer is also the simplest: let AI assist, never decide; add one disclosure line; keep a short record. Get those right and you’ve handled the heart of every AEDT rule — and you can keep the real time-savings AI brings to a small team. The courses above teach exactly where AI belongs in your business and where a human hand still has to stay on the wheel.
See also
Everything below connects to the same underlying skill — using AI responsibly in hiring and work — organized by what you’re trying to do next.
Courses: ChatGPT for Business · AI Fundamentals · Become AI-Fluent · Resume Writing with AI · Study Smarter with AI
Related terms: AI detector · AI hallucination · Agentic AI · Prompt injection · Frontier model
Guides: Using AI to screen résumés? What to tell applicants in 2026 · AI hiring bias 101 for HR teams · AI-generated résumés: how small businesses hire well
Sources
- Automated Employment Decision Tools (Local Law 144) — NYC Department of Consumer and Worker Protection
- Illinois HB 3773 and the Human Rights Act AI amendment — National Law Review
- SB 24-205, Consumer Protections for Artificial Intelligence — Colorado General Assembly
- What the EU AI Act means for hiring and staffing — EU Artificial Intelligence Act
- State AI employment-law compliance updates, 2026 — Seyfarth Shaw
