Check the Interview Process Before AI Checks You

An AI interview is not automatically a reason to abandon an application. The stronger warning is an employer that will not explain the system, what it measures, who reviews the result or how to request another format.

You can often identify those problems before tailoring a résumé or recording a response. Check the job post, careers FAQ, privacy notice and first pages of the application for these seven red flags.

1. The AI interview appears without warning

Look for terms such as automated interview, on-demand video, digital interviewer, conversational AI, automated assessment or AI evaluation. A one-way recording is not necessarily AI-scored, so do not assume that “video interview” alone means an algorithm will evaluate you.

The concern is a process that discloses automation only after you create an account, upload a résumé or begin the interview. In a Greenhouse survey released in May 2026, 70% of 1,200 active U.S. job seekers said they were not clearly told upfront that AI would evaluate them; 21% found out only after the interview started. This was a vendor-published survey, not a population estimate, but it identifies a concrete candidate-experience failure: surprise use of AI (Greenhouse, May 1, 2026).

Before applying, ask: “Will AI record, transcribe, score or recommend a decision at any stage?”

2. Nobody will say what the system measures

“AI-powered insights” is not a useful explanation. A candidate should be able to learn whether the tool evaluates answer content, job-related competencies, keywords, voice transcription, response timing or something else.

Treat claims about analyzing eye contact, facial expression, emotion, personality from appearance or “authenticity” as a higher-risk signal. Even an accurately measured signal may be irrelevant to the job or create barriers for candidates whose speech, movement or interaction differs because of a disability.

The EEOC warns that, without proper safeguards, algorithmic tools can unlawfully screen out people with disabilities who could do the job, and says employers should have a reasonable-accommodation process when using them (EEOC, May 12, 2022).

Before applying, ask: “What inputs and job-related criteria affect the score?” If the answer is “proprietary” or limited to “communication,” you still do not know what behavior the system rewards.

3. Nobody can explain whether AI makes the cutoff

There is a material difference between software that schedules an interview, transcribes answers or organizes notes and software whose score determines who advances.

An automated cutoff gives you less opportunity to correct a bad transcript, explain an unusual career path or challenge a system error. In the Greenhouse survey, 33% identified a prerecorded video interview scored by AI with no human present as a reason to withdraw. Separately, 38% wanted confirmation that a person reviews the AI evaluation before a decision. These are candidate preferences, not proof that human review makes every system reliable.

Before applying, ask:

• Does a person review every interview or only selected ones?
• Can the reviewer see my full answers, not just an AI summary?
• Is the score advisory, or can it reject me automatically?
• Can I correct a materially inaccurate transcript?

“No human interviewer” is not identical to “no human review.” The employer should be able to distinguish them.

4. There is no visible accommodation route

A generic technical-support link is not an accommodation process. Look for a named email address, phone number or form that lets you request a different interview format, extra time, accessible materials, corrected transcription or human review.

Under the U.S. ADA, covered private employers with at least 15 employees and state and local government employers must provide reasonable accommodations that enable qualified applicants with disabilities to compete, unless doing so would cause undue hardship. The EEOC says an applicant may request a change to the hiring process orally or in writing and should give advance notice when possible (EEOC guidance).

The effective alternative depends on the barrier. Extra time will not fix inaccurate speech-to-text. A written response, accessible platform or live interview may. Use this AI video interview accommodation guide to frame a specific request.

5. The privacy explanation does not cover recordings and derived data

A conventional website privacy notice may say little about interview recordings, transcripts, scores, inferred traits or vendor access. Before submitting sensitive data, look for answers to five questions:

1. What will be collected?
2. Who receives it—the employer, interview vendor or additional service providers?
3. How long are the recording, transcript and score retained?
4. Will the data be used to train or improve models?
5. How can you request access or deletion where applicable?

Silence does not prove misuse. It does mean you cannot make an informed tradeoff.

Rules are location-specific. For positions based in Illinois, the Artificial Intelligence Video Interview Act requires notice, an explanation of how the AI works and the general types of characteristics it evaluates, and consent before an employer uses AI to analyze an applicant-submitted video. On request, the employer must delete the applicant’s video interviews within 30 days and instruct recipients of copies to delete them too (Illinois General Assembly, statutory text accessed Sept. 27, 2026). Check current rules for the position’s jurisdiction; requirements change, and legal advice may be needed for a specific case.

6. Monitoring is broader than the job requires

Watch for notices about gaze tracking, browser surveillance, screen recording, keystrokes, room scans or device fingerprinting. Ask what each signal is intended to establish and whether it affects your score.

Identity and assessment integrity can be legitimate goals. That does not make every monitoring method necessary or job-related. An employer should be able to separate identity verification, proctoring and candidate evaluation, rather than presenting them as one vague AI function.

If the role does not require sustained eye contact, for example, the employer should have a better explanation than “the system checks engagement.”

7. There is no person to contact before you commit

A recruiter does not need to conduct every first-round interview. But a candidate should have somewhere to send questions about scoring, accessibility, a malfunction or data use.

Recent candidate accounts show why. CNBC reported an AI interviewer repeating a question and interrupting when a candidate paused, while another system repeatedly prompted a candidate with involuntary eye movement to look straight ahead. Some candidates ended interviews or said they would avoid employers requiring them in future (CNBC, Sept. 15, 2026). These are individual experiences, not proof that every AI interview fails. They do show why a process needs an escalation route when automation goes wrong.

Decide whether to apply, clarify or skip

Apply when the employer identifies the interview format, explains the criteria, describes human oversight and provides clear privacy and accommodation contacts.

Clarify first when the role is attractive but one or two details are missing. Send a short message naming the interview stage, your question and, if relevant, the alternative you need. These questions to ask before a hiring assessment can help.

Skip or withdraw when the employer surprises you with AI, refuses to explain consequential scoring, offers no effective accommodation path or demands intrusive monitoring without a job-related reason. Refusal may end the application, so weigh that cost before deciding; this guide explains what can happen when you refuse an AI interview.

The most useful pre-application test is not “Does this employer use AI?” It is: Can the employer explain where AI acts, what evidence it uses and how a person can correct the process when it fails?