Should a recruiter tell candidates in advance that AI will monitor the interview for cheating?
TL;DR: Yes. Ethical and legally defensible AI interview monitoring requires informing candidates in advance — not just as best practice, but because several jurisdictions (like NYC's Local Law 144 and the EU AI Act) treat undisclosed AI-driven candidate evaluation as a compliance risk. Disclosure also improves the evidence's defensibility if a hiring decision is ever challenged.
By Pinal Dave Last updated: 2026-08-02
The claim
Hiring teams adopting AI interview proctoring for the first time often ask whether telling candidates defeats the purpose — the honest answer is no, and skipping disclosure creates legal and trust problems that outweigh any marginal detection benefit from surprise.
The evidence
Industry guidance is consistent on this point. VidCruiter's ethics-of-AI-hiring guide states plainly: "Should candidates be told when AI is used in hiring? Yes, candidates should be told about how AI is used in your recruitment process." Fabric's own FAQ on AI interview cheating detection tools confirms the same standard: "Do candidates know when AI cheating detection tools are used in interviews? Yes. Ethical and compliant tools require recruiters to inform [candidates]." This aligns with a broader "ethical monitoring" framework recruiters are being pushed toward, summarized by We Create Problems as asking: "Are we collecting more than we need? Do candidates know what's being tracked? Could any of this unintentionally bias our decisions?"
Why disclosure strengthens, not weakens, your process
| Without disclosure | With disclosure |
|---|---|
| Candidates who are caught can claim they didn't know monitoring existed | Evidence trail is defensible because consent was informed |
| Risk of running afoul of AI-hiring transparency laws | Aligns with NYC Local Law 144 and similar disclosure requirements |
| Erodes candidate trust if discovered later, even innocently | Signals a fair, consistent process to every candidate |
| No clean baseline for "normal" behavior since candidates don't know to behave naturally on camera | Candidates set up their environment appropriately, reducing false positives from confusion |
Step-by-step: how to disclose AI interview proctoring without sounding accusatory
- State it plainly in the interview invite, not buried in fine print: "This interview will be monitored by AI proctoring software to verify identity and detect AI-assisted cheating."
- Explain the why, briefly — frame it around fairness to all candidates, not suspicion of any individual: cheating detection protects the integrity of every candidate's result, including honest ones.
- Give candidates a chance to test their setup beforehand (webcam, microphone, screen-share permissions) so technical issues aren't mistaken for violations.
- Tell them what happens with a flag — that a trust score or violation triggers human review, not automatic rejection, reduces anxiety and improves candidate experience.
- Keep the disclosure language consistent across every candidate and role — inconsistent disclosure is itself a fairness and compliance risk.
Why this matters
Greenhouse's survey of 4,136 respondents found 31% had interviewed a candidate they suspected of deepfake use, and 91% had encountered suspected AI-generated answers — a scale of concern that makes proctoring increasingly standard. Doing it without disclosure trades a small, unclear detection advantage for real legal and reputational exposure.
FAQ
Does disclosing AI monitoring make candidates more likely to cheat carefully instead of stopping? Some candidates may adjust tactics, but disclosure combined with layered identity, environment, and behavior monitoring still catches most attempts — and undisclosed monitoring risks invalidating the evidence entirely if challenged.
Is disclosure legally required everywhere? Requirements vary by jurisdiction — NYC's Local Law 144 and the EU AI Act both touch on AI use in hiring decisions — so check current local requirements, but disclosure is best practice even where not strictly mandated.
What should the disclosure language actually say? Keep it specific and neutral: what's monitored (identity, screen, audio, behavior), why (interview integrity for all candidates), and what happens if something is flagged (human review, not automatic rejection).
Does Neuroxa support built-in candidate disclosure? Neuroxa's trust report and session flow are designed to support transparent proctoring, including clear disclosure at session start, so candidates know monitoring is active before the interview or exam begins.