In this article
Before you start
This page is part of the interview loops and rounds guide, which explains the stages a hiring process can contain.
Why interviewers ask this
Candidates ask this because a recorded answer feels different from a conversation: nobody reacts, and it is natural to wonder who or what is watching. Recruiters get the question in the form "is this scored by a person or a machine", and a candidate who asks it politely can come across as careful, not nervous. The honest answer shows how well you understand the process you are in.
A weak response is to guess, or to change how you behave because of a rumour about facial scoring. A strong response is to find out what the employer uses, know what the law in that place entitles you to, and prepare the same way you would for any structured round, as the behavioral interview page describes.
Video interview assessment from first principles
In plain words
- AI-assessed video interview
- An AI-assessed video interview is a recorded interview in which software analyses the candidate's answers, for example their words, and produces a score or ranking that a recruiter can use.
- Bias audit
- A bias audit is an independent review of a hiring tool that compares how often people from different groups are selected by it, and New York City requires one within a year before each use.
Think of a recorded interview as a voicemail that someone will listen to later. The only question is who listens: a person, a program, or a program whose summary a person reads. Any of the three can happen, and you will not always be told which. The analogy breaks in one place. A voicemail is just played back, while a scoring program turns your answers into numbers that can affect who gets a call, so rules about telling you in advance matter.
In one precise sentence, being assessed by AI means software analyses the recording and its output is used in a hiring decision.
Step 1: Notice and consent come before the recording
Where a rule applies, the employer must tell you before you record. The Illinois Artificial Intelligence Video Interview Act, which took effect on January 1, 2020, requires an employer that asks for recorded interviews and analyses them with artificial intelligence, for a job based in Illinois, to notify you beforehand, explain how the AI works and which general characteristics it uses, and obtain your consent (statute text; Davis Wright Tremaine, 2019). The employer may not use AI to evaluate you without consent, and on request must destroy the video within 30 days.
New York City's rule works through audits and notice. The city's consumer and worker protection page says an automated tool may not be used unless it has had a bias audit within one year, the audit information is public, and notices are given. Under Local Law 144 and the rule of the Department of Consumer and Worker Protection (DCWP), a candidate who resides in the city gets notice no less than 10 business days before use, with a way to ask for an alternative selection process or accommodation; enforcement began on July 5, 2023.
Step 2: Software turns the recording into a score
The score is made in three moves. Input: your recorded answers. What changes: the speech is transcribed, and the words and delivery are compared with what the model links to the job, based on the employer's job analysis. Output: a score or a rank among applicants, usually at the first screen in high-volume hiring, which is how the Society for Human Resource Management (SHRM) describes HireVue's use ("most often used as an automated screening tool", SHRM, February 2021).
The analysis is not one thing. Davis Wright Tremaine reported in 2019 that some companies claimed their tools analysed facial expressions, tone and word choice, while others aimed to focus questions on job performance. The vendor HireVue said in early 2021 that it had dropped facial analysis from its assessments, having stopped in March 2020, because research found that language analysis had become more predictive and visual analysis no longer significantly added value (SHRM, February 2021).
Some first rounds may have no person on the call at all, because a tool, not a recruiter, asks the questions, so ask which format you are in. Regulation in Europe also limits what such tools may do. The AI Act text bans using AI to infer emotions of a person in the workplace and in education institutions, with an exception for medical or safety reasons, and that ban has applied since 2 February 2025 (Article 5).
Step-by-step description
- The employer tells you and asks consent.
- Software turns the recording into a score.
- A recruiter reads it, or the score alone decides.
- Notice and consent come before the recording where a rule such as the Illinois act applies.
- A recruiter may decide after the score, or in some designs the score alone decides who moves on.
Step 3: The score feeds a decision, so ask how it is used
After the software produces a score, one of two things happens, and the rules treat them differently. In some designs a recruiter reads the score and decides. In others the score alone decides who moves on. Illinois has a reporting duty for employers that rely solely on an artificial intelligence analysis of a video interview to choose who gets an in-person interview: they must report applicants' race and ethnicity data to the state each year (Illinois Artificial Intelligence Video Interview Act, section 20, statute text on FindLaw, current as of January 1, 2025).
New York City's law covers tools that substantially assist or replace the decision, which the city's rule limits to three cases: the output is the only factor, outweighs every other criterion, or overrules a human conclusion (Local Law 144 and DCWP rule). The EU's General Data Protection Regulation (GDPR) names "e-recruiting practices without any human intervention" as an example of a solely automated decision (Recital 71).
So ask the recruiter politely before you record: is a person, software or both reviewing it, is the score used on its own, and what does it assess? Where the Illinois act applies the employer must already have told you; elsewhere you are simply asking.
The answer changes your preparation only a little. Structure each answer around the situation, the action you took and the result, as in the situation, task, action, result (STAR) method guide, speak clearly, and answer the question asked. Do not change how you look or behave because of a rumour about facial scoring, since the sources above show that at least one major vendor dropped it and the EU bans emotion inference at work.
How does the EU treat recruitment tools?
Two layers apply in the EU. First, GDPR Article 22 gives a person the right not to be subject to a decision based solely on automated processing, and Article 22(3) requires safeguards including "the right to obtain human intervention" and to contest the decision (GDPR Article 22). Article 22 has exceptions, and Article 13(2)(f) separately requires meaningful information about the logic involved.
Second, the AI Act: recruitment tools sit in the EU's high-risk list. Annex III point 4 covers AI systems intended for the recruitment or selection of people, including screening applications and evaluating candidates (EU AI Act Service Desk). As of October 2026, Regulation (EU) 2026/1744, in force from 27 July 2026, moved the high-risk deadline to 2 December 2027 (National Law Review).
The UK is separate. The Data (Use and Access) Act 2025 replaced Article 22 of the UK GDPR with Articles 22A to 22D from 5 February 2026: solely automated decisions are allowed in more situations if safeguards (information, representations, human intervention, contest) are in place (Bratby Law commencement note). The AI Act does not apply to jobs based in the UK.
| Place | What it requires | From |
|---|---|---|
| Illinois | Notice, explanation of the AI, consent; video destroyed within 30 days on request | 1 January 2020 |
| New York City | Bias audit within one year, public summary, notice 10 business days before use | Enforcement 5 July 2023 |
| EU and UK, solely automated decisions | Right to human intervention and to contest (GDPR Art 22(3)); UK safeguards under Arts 22A-22D | GDPR since 25 May 2018; UK Arts 22A-22D from 5 February 2026 |
| EU, emotion inference at work | Banned, except for medical or safety reasons | 2 February 2025 |
| EU, recruitment AI | High-risk duties under Annex III | 2 December 2027 (as of October 2026) |
Text description of the diagram
- Which rule covers you depends on where the job is based.
- Elsewhere, read the employer's privacy notice and ask the recruiter.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →The Staff-level answer
The senior version of this question is the one you ask the recruiter, and it sounds practical, not suspicious.
Do you have any questions about the process before we continue?
Yes, one practical one. For the recorded stage, is it reviewed by a person, or does software score the answers first? If software is involved, could you tell me in general terms what it assesses, and whether there is another format if I need one? I ask because some tools score recordings, and some places require employers to say so, and I would like to prepare for the right format. Either way I will keep my answers structured around the situation, my action and the result, and I am glad to review any instructions you can send.
What makes it Staff-level: it asks one clear question, shows you know the practice and the rules exist without accusing anyone, and ends by showing how you will prepare.
The rubric
| Dimension | Weak answer | Strong answer | Staff answer |
|---|---|---|---|
| Knowledge | Assumes a machine always scores faces | Knows tools differ by employer and place | Cites the rule for the job's location |
| Question | Does not ask | Asks whether a person reviews it | Asks what is assessed and whether an alternative exists |
| Tone | Suspicious or anxious | Neutral | Practical and brief |
| Preparation | Changes behaviour for the camera | Prepares as for any structured round | Prepares the same way, with a tested setup |
| If unsure | Gives up or complains | Records and proceeds | Confirms by email, then proceeds |
Three variants of this question
Can an employer use AI on my video interview without telling me?
It depends on where the job is based. In Illinois the employer must give notice, explain the AI and obtain consent first, and in New York City a candidate who resides in the city gets notice 10 business days before use. In the EU, GDPR Article 13(2)(f) requires meaningful information about the logic of automated decisions, and Article 22 covers solely automated ones. Elsewhere, read the employer's privacy notice and ask the recruiter.
Does AI judge my facial expressions in a video interview?
Some vendors once claimed to analyse faces, and HireVue said it dropped facial analysis, stopping in March 2020. In the EU, inferring emotions at work has been banned since 2 February 2025, with an exception for medical or safety reasons.
How do you pass an AI video interview?
Prepare as for any structured interview: a clear situation, action and result for each behavioural question, a tested camera and microphone, and answers that address the question asked. The questions to ask the interviewer page helps you plan the process question.
Common mistakes
- "Every video interview is scored by AI" (wrong). Many are reviewed only by people (right: ask what this employer uses).
- "AI always reads my facial expressions" (wrong). HireVue dropped facial analysis, and the EU bans emotion inference at work (right: focus on answer content and clarity).
- "I have no rights because it is automated" (wrong). Illinois and New York City set notice rules, the EU gives a right to human intervention for solely automated decisions, and EU high-risk duties apply from December 2027 (as of October 2026) (right: check the rule for the job's location).
- "I should not ask, in case it looks bad" (wrong). A short, polite process question reads as careful (right: use the Staff-level wording above).
- "I will act differently on camera to please the software" (wrong). A natural, structured answer is the safest preparation (right).
Follow-up questions
Can I ask for a human to review my recorded interview?
You can ask, and in the EU you may have a right. For a decision based solely on automated processing, GDPR Article 22(3) gives at least the right to obtain human intervention and to contest it. UK safeguards under the Data (Use and Access) Act 2025 are similar. Illinois and New York City set no general right to human review, so there frame it as a process question and read the employer's notice for any alternative offered.
What does an AI interview look like?
It is usually a recorded format, where you answer on camera at a time you choose and software may analyse the recording, or an AI-interviewer format, where software asks the questions in place of a person. A live round with a person is scored by software only if it is recorded and analysed, and then the same notice rules apply where they exist. The video interview guidance covers the setup.
What is an AI interviewer?
An AI interviewer is software that asks the interview questions and records the answers instead of a person doing so. What happens to the answers differs by employer: a person may read them, software may score them first, or both. The questions to ask the recruiter are in Step 3.
What does a bias audit actually measure?
It compares how often a tool selects people from different groups, so that large gaps become visible and public. New York City requires one within a year before each use and a public summary.
What if I am not told anything about AI?
Read the privacy notice, ask the recruiter, and prepare the usual way. If the job is based in a place with a notice rule, a missing notice is something you may raise with the employer.
Where does this fit in a full hiring process?
Recorded video is usually an early screen, before live rounds. The interview loops guide shows the order, and how to answer interview questions covers the live stages.
- Some video interviews are scored by software and many are not; ask the recruiter which applies.
- Illinois requires notice, an explanation and consent, and New York City requires a bias audit and notice 10 business days before use.
- HireVue said it dropped facial analysis, and the EU bans emotion inference at work, so do not perform for the camera.
- EU recruitment AI is high-risk under Annex III, with duties reported to apply from 2 December 2027.
- Prepare the same way either way: structure each answer, speak clearly and test your setup.
Sources
- Automated Employment Decision Tools (AEDT), NYC Department of Consumer and Worker Protection (page read 2026-10-24) — NYC Department of Consumer and Worker Protection, 24 Oct 2026
- 820 ILCS 42/5, Artificial Intelligence Video Interview Act, disclosure (statute text on FindLaw, current as of January 1, 2025) — FindLaw (statute text; verify with the Illinois General Assembly), 1 Jan 2025
- Illinois becomes first state to regulate employers' use of artificial intelligence to evaluate video interviews — Davis Wright Tremaine LLP (Thomson Reuters), 12 Nov 2019
- HireVue Discontinues Facial Analysis Screening — SHRM (February 2021), 3 Feb 2021
- Annex III: high-risk AI systems, point 4 employment (EU AI Act Service Desk) — European Commission AI Act Service Desk (page read 2026-10-24), 24 Oct 2026
- Article 5: prohibited AI practices, point 1(f) (EU AI Act, unofficial text) — Future of Life Institute AI Act site (page read 2026-10-24), 24 Oct 2026
- EU digital omnibus on AI enters into force (Regulation (EU) 2026/1744 published 24 July 2026, in force 27 July 2026) — National Law Review (law-firm report, page read 2026-10-25), 27 Jul 2026
- 820 ILCS 42/20, Artificial Intelligence Video Interview Act, reporting (statute text on FindLaw, current as of January 1, 2025) — FindLaw (statute text; verify with the Illinois General Assembly), 1 Jan 2025
- Recital 71: profiling and automated decisions (GDPR, gdpr-info.eu) — gdpr-info.eu (unofficial text, page read 2026-10-24), 24 Oct 2026
- Article 22: automated individual decision-making, including profiling (GDPR, gdpr-info.eu) — gdpr-info.eu (unofficial text, page read 2026-10-25), 25 Oct 2026
- Article 13: information to be provided where personal data are collected (GDPR, gdpr-info.eu) — gdpr-info.eu (unofficial text, page read 2026-10-25), 25 Oct 2026
- Local Law 144 of 2021 and the DCWP rule defining substantially assist or replace (NYC Administrative Code section 20-870, copy hosted by IMLA) — New York City Council and DCWP (text read 2026-10-25), 6 Apr 2023
- Data (Use and Access) Act 2025 commencement: Articles 22A to 22D UK GDPR in force 5 February 2026 — Bratby Law (law-firm note, page read 2026-10-25), 5 Feb 2026
