What is an AI interviewer? How it works, where it is used and the rules that apply

An AI interviewer is software that conducts a job interview itself, most often the first-round screen. It asks questions by voice, video or chat, adapts follow-up questions to the candidate's answers, and produces a transcript, a summary and usually a score against criteria the employer set, which a recruiter reviews before anyone advances. Examples on vendors' own sites as of October 2026 include BrightHire Screen, Greenhouse Voice AI and HireVue's announced AI Interviewer. Because it evaluates candidates rather than just recording a human-led conversation, AI hiring laws such as New York City's Local Law 144, Illinois's AI video interview law, California's automated decisionmaking rules and the EU AI Act apply more directly than they do to note takers.

How does an AI interviewer work?

The vendors describe broadly the same sequence. Greenhouse's glossary entry on AI interview agents sets it out in four steps:

1. Set-up. The hiring team defines the questions and the scoring rubric for the role.
2. The interview. The candidate joins on their own schedule, by phone, browser or app, and the agent asks the questions, choosing follow-ups based on what the candidate says.
3. Scoring. The agent scores the answers against the rubric and produces a transcript and a rationale.
4. Human review. A recruiter reviews the evidence and decides who moves forward.

Greenhouse distinguishes an "agentic" interview, which chooses the next question based on the previous answer, from a chatbot that asks scripted questions in a fixed order. That adaptive follow-up is what separates an AI interviewer from older tools.

How is it different from a one-way video interview, a chatbot or an AI note taker?

Interview Signal, which we make, belongs in the last two rows: it assists the person running the interview and does not conduct interviews. The wider vocabulary is in the AI recruiting glossary and what is interview intelligence.

What do AI interviewers on the market actually do?

Three examples, described from each vendor's own pages as of October 2026:

The vendors' own positioning is consistent on one point: the agent surfaces evidence and a person decides. Whether that is true in your process depends on how you configure and use the tool, which is also what decides how the law treats it.

What are AI interviewers good for, and where do they fall short?

Where they help:

• Volume. High-volume roles where recruiters cannot phone-screen every qualified applicant.
• Availability. Candidates interview at any hour, which suits shift workers and other time zones.
• Consistency. Every candidate gets the same core questions and is scored against the same rubric.
• A record. Transcripts make it possible to check why someone scored as they did.

Where they fall short:

• Candidate experience. Some candidates will not talk to software, and a screen is often their first conversation with you.
• Accessibility. Candidates with speech, hearing or processing differences can be disadvantaged by a voice agent. The ADA requires tests to measure the skill, not "impaired sensory, manual, or speaking skills" (42 U.S.C. § 12112(b)(7)), so an alternative route is needed; see interview accommodations for neurodivergent candidates.
• Validity. A score is only as good as the rubric and the evidence that it predicts performance.
• Integrity. Asynchronous interviews give candidates room to read AI-generated answers; follow-up quality matters.
• Selling the role. Senior or scarce candidates expect a person who can answer questions and persuade them.

What laws apply to AI interviewers?

Because an AI interviewer evaluates candidates, it sits on the stricter side of most AI hiring rules. As of October 2026, not legal advice:

What should you ask a vendor before using an AI interviewer?

• Who writes the questions and the rubric, and can we approve every question before it is asked?
• What does the score measure, and what evidence shows it predicts performance in roles like ours?
• Does it analyze voice, tone, accent or facial expression, or only the content of answers?
• Can it advance or reject anyone automatically, and can that be switched off?
• What bias testing has been done, by whom and when? Is an independent audit available for New York City use?
• How does a candidate request a human interview or an accommodation, and what happens to their application?
• How does the agent disclose that it is AI, and what does it tell candidates about data use?
• Where are recordings and transcripts stored, for how long, and are they used to train models?
• Can you explain one named candidate's score, in a form we could send them?

How should you tell candidates?

Tell them before they start, in the invitation, and make the alternative real. A line you can adapt:

Your first interview for this role is with an AI interviewer, software
that asks you [number] questions by [voice / video / chat] and may ask
follow-up questions. It takes about [x] minutes and you can do it at
any time before [date]. A recruiter reviews your answers and decides
who moves forward; the AI does not make that decision.
If you would prefer to speak with a person, or need an accommodation,
reply to this email. Choosing a person does not affect your
application. How we use your information: [link].

Fuller wording for each stage, and the laws behind it, is in the AI disclosure templates.

When should a person do the interview instead?

AI interviewers fit high-volume, well-defined roles where the first screen checks a stable set of requirements. Keep a human interviewer when the role is senior or scarce and the call is partly a sales conversation, when the requirements are ambiguous and need judgment to probe, when a candidate asks for a person, and for any stage where the decision is final. Many teams use both: the AI interviewer for the first screen, and people, supported by notes and scorecards, for everything after.