AI Hiring Rules and Rights

Hiring software can store applications, search for keywords, apply knockout rules, rank people, score assessments, or recommend a decision. Those functions are not interchangeable, and neither are the laws that may apply to them.

Use this guide as a route into the evidence. Start with your role, then check the hiring stage, system function, and jurisdiction before relying on a rule.

If You Are a Candidate

Before an automated interview or assessment

• Ask what platform and format will be used.
• Request an accommodation from the employer when a disability makes the standard format inaccessible. A request does not need special legal wording.
• Do not assume every recorded interview uses facial analysis, voice scoring, or automated ranking; product configuration matters.

Understand what AI video interviews may score, then use the accommodation request guide for a step-by-step path.

After an automated hiring decision

The information you can request depends on where the job, employer, and candidate are covered. New York City Local Law 144 provides notice and general information rights for covered automated employment decision tools, but it does not promise an individual score or a causal explanation for a rejection.

Read the New York City Local Law 144 guide, then see what you can request after an AI-assisted rejection. The second guide separates voluntary questions from rights that exist only under particular laws.

While preparing an application

Most applicant tracking systems are databases first. Keyword search, resume parsing, knockout questions, and machine-learning rankings are separate mechanisms.

• Does AI read your resume?
• Write a resume that survives parsing and screening

If You Are an Employer or HR Team

Identify the system’s real function

Document whether the tool stores, searches, filters, ranks, scores, recommends, or decides. A product label such as “AI recruiting” does not establish how a configured workflow operates or whether a particular legal definition is met.

Map what AI actually automates across HR.

Test outcomes without overstating the result

Selection-rate and impact-ratio checks can reveal an important disparity. They do not by themselves establish individual fairness, job validity, accessibility, or a conclusive legal violation. Position-level differences can also disappear when unlike jobs are pooled together.

See how hiring bias enters a system and what an audit can test.

Track obligations by jurisdiction and date

• New York City: Covered AEDTs require a recent bias audit, a public summary, and advance notice. Enforcement began in July 2023.
• European Union: Employment remains an Annex III high-risk area. Some transparency and prohibited-practice rules already apply, while the amended high-risk timetable sets 2 December 2027 for stand-alone Annex III systems. Classification guidance and standards remain moving parts.
• Other US states: Terminology, covered decisions, notice, access, and human-review rights differ. Do not reuse an NYC definition as a national one.

Read the grounded EU timeline. It separates the official dates from system-specific questions that still require checking against the operative law.

Evidence Rules for This Library

1. Prefer statutes, regulations, agencies, courts, and official technical material for legal and procedural claims.
2. Label vendor performance and fairness statements as vendor claims unless independently tested.
3. Date every changing rule or product claim.
4. Keep contested allegations and unresolved questions visibly unresolved.
5. Treat HRaizon articles as coverage—not as independent proof of their own claims.

This page is informational, not legal advice. Confirm obligations for the actual system, decision, people, and jurisdictions involved.