For employers, HR & talent ops

Could you defend your AI hiring toolif a regulator asked tomorrow?

Automated employment decision tools now sit inside a patchwork that is in force — NYC bias audits, Illinois notice rules, California recordkeeping. Score the evidence dossier for each tool you deploy and get one verdict — DOSSIER READY AS DESCRIBED, SHORTFALL, or EVIDENCE GAP — with a gate that won't let a tool pass on paperwork while the load-bearing controls are missing.

Get the Dossier — $99one-time · instant download · yours to keep

Not legal advice. This is a readiness aid that grades a deployer's evidence dossier from your own marks — not a bias audit, certification, opinion of counsel, or safe harbor. AEDT obligations vary by jurisdiction and change often; it scores no candidate or employee. Confirm your specific obligations with a qualified employment attorney.

Five deliverables · runnable
Runnable scoring engine
Python
Workbook that reproduces it
.xlsx
Dossier-assembly playbook
.docx
Gap-closure runbook
.docx
6-tool worked sample
.csv
Works alongside
Recruiting Skills Pack · Resume-Claim Gate · Risk Triage

Update notice — version 1.4.0 (2026-11-15). This package replaces version 1.3.0 of the AEDT Deployer Compliance Dossier. What did not change: the six controls, their weights and their 0, 1 and 2 mark texts, the band thresholds (75 and 50), the two gate triggers (independent bias audit marked 0, or pre-use candidate notice marked 0), the result names (DOSSIER READY AS DESCRIBED, SHORTFALL, EVIDENCE GAP) and the portfolio results, and the score of every combination of 0, 1 and 2 marks. The order in which a control is named is also unchanged: lowest mark first, with ties going to a gate control, then to the heavier weight, then to control order. What changed: New top-band rule: a zero on any control prevents the top result. A tool whose score is 75 or more, with no gate fired and any control marked 0, now reads SHORTFALL instead of DOSSIER READY AS DESCRIBED, and the result names each control at 0. Of the 729 possible combinations of 0, 1 and 2 marks for one tool, 20 move from DOSSIER READY AS DESCRIBED to SHORTFALL, and no other combination changes its result. The demo, the engine and the workbook print a sentence naming the control at 0. New answer, Not applicable: Human review and appeal, recordkeeping, and data-source and proxy-governance controls may be marked Not applicable only when the user’s documented applicability review concludes that the control does not apply to the tool and use being assessed. Not applicable records the user’s assessment—not RedHub’s legal determination—and excludes that control from the score. A control marked NA is left out of the score, both the points it earns and the points it could earn. It never clears a gate, never counts as a zero under the top-band rule and is never named as the priority control. A 0 now means missing evidence only. The two gate controls and audit publication still take 0, 1 or 2. Priority control: the "Close first" line and column are now "Priority control". The line is never blank: it names a control whenever any mark is below 2, including at DOSSIER READY AS DESCRIBED, and otherwise reads "No scored control is below the top mark." Re-run earlier rows: yes. A tool scored in an earlier version at 75 or more, with neither gate control at 0 and a 0 on audit publication, human review and appeal, recordkeeping, or data-source and proxy governance, read DOSSIER READY AS DESCRIBED. In version 1.4.0 it reads SHORTFALL. Every other combination of 0, 1 and 2 marks keeps its score and its result. If an earlier row marked human review and appeal, recordkeeping, or data-source and proxy governance 0 and the user’s documented applicability review concludes that the control does not apply to the tool and use being assessed, re-mark it NA and re-run the row. Its score and its result can change.

01.The Problem

The dashboard says you're compliant. The file says otherwise.

$500–$1,500

NYC penalties run up to $500 for a first violation and $500–$1,500 for each subsequent violation, accruing per day. A December 2025 New York State Comptroller audit found DCWP falling short in enforcing the law: its complaint intake was ineffective, and auditors reviewing the same companies identified potential non-compliance DCWP’s own review had missed. DCWP disagreed with some findings and agreed to adopt most of the recommendations in part or in full.

self-attestation

a gate-triggering gap to check for: a vendor's own “bias report” passed off as the audit. For NYC Local Law 144, the required audit must be conducted by an auditor who satisfies NYC’s independence criteria. A vendor’s internal self-assessment is not, by itself, that independent audit — though a vendor may arrange for a qualifying independent auditor to audit the tool.

patchwork

NYC, Illinois, California, soon Connecticut — each with different notice, audit, and recordkeeping rules. A tool defensible in one place can be exposed in another.

This grades the evidence you can actually produce against a composite US readiness baseline. If the independent-bias-audit evidence control is marked 0 or the pre-use-notice evidence control is marked 0, this instrument changes the final result to EVIDENCE GAP regardless of score. These are user-supplied marks; RedHub did not determine whether an audit or notice was required or whether any evidence satisfies an applicable requirement. These controls combine selected obligations and governance practices drawn from multiple US jurisdictions. Not every control is legally required for every tool, employer, location, or employment decision. Confirm scope and applicable requirements before treating a missing control as legally required.

02.See It Work

Score a tool and watch the gate do its job.

Before running

I am authorized to enter and share information about this organization and understand that the result is not an independent assessment by RedHub.

Diagram of the AEDT Deployer Compliance Dossier: six weighted controls, a two-trigger gate, and a sample scoring 76 that still reads EVIDENCE GAP.
Shareable diagram

How the gate works, in one image

The six weighted controls, the band thresholds, and the two-trigger gate that forces EVIDENCE GAP — the same logic the demo runs, as a diagram you can share or embed anywhere.

View & embed the full diagram
03.The Runnable Engine

One command, every tool, an auditable verdict.

The zero-dependency Python engine reads your tool list and prints the same verdict the workbook and demo produce. The Resume screener below scores 76 and still reads EVIDENCE GAP — the only audit on file is the vendor's own.

AEDT Deployer Compliance Dossier
====================================================
Before running, confirm: I am authorized to enter
and share information about this organization and
understand that the result is not an independent
assessment by RedHub.
----------------------------------------------------
Resume screener (NYC)       76/100  EVIDENCE GAP     [GATE -> EVIDENCE GAP]
    Priority control: Independent bias audit on file (within the last year)
Video-interview analyzer    52/100  EVIDENCE GAP     [GATE -> EVIDENCE GAP]
    Priority control: Pre-use candidate notice delivered
Promotion-scoring model     94/100  DOSSIER READY AS DESCRIBED
    Priority control: Data-source & proxy-discrimination governance
Candidate-matching engine   50/100  SHORTFALL      
    Priority control: Independent bias audit on file (within the last year)
Chatbot assessment          48/100  EVIDENCE GAP   
    Priority control: Audit-publication requirement assessed
Internal mobility scorer   100/100  DOSSIER READY AS DESCRIBED
    No scored control is below the top mark.
----------------------------------------------------
Portfolio: REVIEW BEFORE CONTINUED USE
3 of 6 tool(s) read EVIDENCE GAP.
One or more tools triggered this instrument’s
evidence gate. Determine whether the missing control
is applicable and whether continued use is
appropriate before relying on the portfolio result.
This result does not itself direct suspension or
establish a legal violation.
----------------------------------------------------
Every result above:
This result reflects marks supplied by the user.
RedHub did not inspect the tool, audit, notices,
data, deployment, or legal obligations. Do not
present this result as a certification, bias audit,
legal opinion, or independent third-party assurance
report.
04.The Standard

Six controls, weighted to 100 — two of them gates.

Independent bias audit on fileGate
24
Pre-use candidate notice deliveredGate
20
Audit-publication requirement assessed
16
Human review & appeal path
14
Recordkeeping & retention
14
Data-source & proxy governance
12
The gate does distinct work

If the independent-bias-audit evidence control is marked 0 or the pre-use-notice evidence control is marked 0, this instrument changes the final result to EVIDENCE GAP regardless of score. These are user-supplied marks; RedHub did not determine whether an audit or notice was required or whether any evidence satisfies an applicable requirement. That does not establish that the control is legally required or that the deployment violates applicable law. The gate only worsens a verdict, never lifts one.

Composite US readiness baseline

These controls combine selected obligations and governance practices drawn from multiple US jurisdictions. Not every control is legally required for every tool, employer, location, or employment decision. Confirm scope and applicable requirements before treating a missing control as legally required. Human review and appeal, recordkeeping, and data-source and proxy-governance controls may be marked Not applicable only when the user’s documented applicability review concludes that the control does not apply to the tool and use being assessed. Not applicable records the user’s assessment—not RedHub’s legal determination—and excludes that control from the score.

It releases when you fix it

Close a gate control and the gate releases — the verdict returns to whatever the score earned. Whenever a control is below 2, a priority control is named.

05.What This Is — And Isn't

A readiness aid, not a bias audit or a safe harbor.

What it is
  • A deterministic dossier-readiness verdict from your own marks.
  • A way to find a gate-triggering gap before a regulator or candidate does.
  • A portfolio rollup that, where a tool is short, names the worst one to act on first.
  • Offline — engine, workbook, and demo agree to the verdict.
What it isn't
  • Not a bias audit, certification, opinion of counsel, or safe harbor.
  • Not connected to your hiring tool — it runs no audit and reads nothing live.
  • Not a scoring of any candidate or employee — it grades the deployer's file.
  • Not a substitute for an employment attorney on your specific obligations.

Not legal advice. This is a readiness aid that grades a deployer's evidence dossier from your own marks — not a bias audit, certification, opinion of counsel, or safe harbor. AEDT obligations vary by jurisdiction and change often; it scores no candidate or employee. Confirm your specific obligations with a qualified employment attorney.

06.Who It's For

Anyone who has to stand behind an AI hiring tool.

Employers using resume screeners, rankers, or interview AI
HR and talent-ops leaders who own the compliance file
Employment agencies deploying AEDTs for clients
Multi-state employers juggling NYC, Illinois, and California rules
Compliance and legal-ops teams inventorying AI hiring tools
Consultants running an AEDT readiness review for a client
08.Common Questions

The honest answers.

The evidence dossier you, the deployer, can produce for an automated employment decision tool (AEDT) used in hiring or promotion — the independent bias audit, the candidate notices, the published summary, the human-review and appeal path, the recordkeeping, and the data-source / proxy governance. It returns DOSSIER READY AS DESCRIBED, SHORTFALL, or EVIDENCE GAP per tool. A 0 on any control prevents DOSSIER READY AS DESCRIBED: a tool that scores 75 or more with a control at 0 reads SHORTFALL. Whenever a control is below 2, the result names a priority control. Human review and appeal, recordkeeping, and data-source and proxy-governance controls may be marked Not applicable only when the user’s documented applicability review concludes that the control does not apply to the tool and use being assessed. Not applicable records the user’s assessment—not RedHub’s legal determination—and excludes that control from the score. A 0 means missing evidence only. It scores the file, never a candidate or employee.

Find the gate-triggering gap
before they do.

One purchase, lifetime access, 12 months of updates. $99, once.

Not legal advice. This is a readiness aid that grades a deployer's evidence dossier from your own marks — not a bias audit, certification, opinion of counsel, or safe harbor. AEDT obligations vary by jurisdiction and change often; it scores no candidate or employee. Confirm your specific obligations with a qualified employment attorney.

Sold by RedHub AI LLC · Secured by Stripe · redhub.ai