When candidates interview without a human in the room, the question every employer asks is: how do I know this is their own thinking? Here's exactly what integrity monitoring detects — and what it doesn't.
One-way AI interviews give candidates flexibility — they can interview at home, at their own pace, on their schedule. That's the point. But flexibility creates a temptation: the candidate is alone, unobserved, with access to the internet, AI tools, and prepared scripts.
If you're evaluating candidates on the content of their answers, and some candidates are reading answers written by someone else, your scoring data is corrupted. You're not comparing candidates — you're comparing their preparation strategies.
Integrity monitoring exists to flag the signals that suggest a candidate is not engaging authentically with the interview process. It doesn't replace human judgment — it informs it.
Monitors whether the candidate is oriented toward their screen. Repeated extended off-screen gaze may suggest they're reading from a prepared document.
Detects when the candidate navigates away from the interview window — to a browser tab, another app, or an AI tool. Each instance is logged with the question number.
Unusually fast responses with no pause may indicate a pre-written answer being read. Unusually slow responses with subsequent fluency may also warrant review.
Scripted answers often have a distinct cadence — no hesitation, no correction, no natural thinking-out-loud. The AI notes unusual fluency for structured screening questions.
When Sofia asks follow-up questions on a candidate's answer, a genuine answer is easy to expand. A read answer often breaks down under probing — the candidate can't elaborate naturally.
Clarity matters here. HireCredible does not use facial recognition. It does not analyse emotional state or classify emotional expressions. It does not score candidates based on their appearance, background, or any physical characteristic visible on camera.
The camera is used for gaze direction analysis only — a coarse signal about whether the candidate is looking at their screen. This is distinct from facial recognition, which identifies or characterises individuals based on biometric facial features.
HireCredible also does not use any scoring based on accent, voice tone, or speech characteristics beyond content. The scoring model evaluates what was said, not how it was delivered.
See integrity monitoring in action — run a free trial role on AI interview software Singapore teams use.
Start Free →Integrity signals are presented in the candidate's result report alongside their scores. You see the overall score, per-dimension scores, evidence quotes — and an integrity section that lists any flags raised during the session, with context about when they occurred and how many times.
Flags are informational. They surface for your consideration; they do not automatically disqualify a candidate. A single tab switch on Question 1 is different from five tab switches across the entire interview. You make the call.
This keeps human judgment in the loop where it matters — not in the tedious work of conducting the same phone screen 40 times, but in the meaningful evaluation of borderline cases.
HireCredible is transparent with candidates about integrity monitoring. Before the interview begins, the consent screen explains that the session includes integrity signals — gaze monitoring, tab detection, and coherence analysis. Candidates must confirm they understand and agree before proceeding.
This transparency serves two purposes. First, it's the right approach — candidates deserve to know how they're being evaluated. Second, it acts as a deterrent. Knowing that tab switching is logged reduces the temptation to do it, which means the candidates who proceed are more likely to engage genuinely.
What is AI interview integrity monitoring?
Integrity monitoring tracks candidate behaviours during an AI interview that may indicate they are not engaging authentically — such as reading scripted answers, using external assistance, or switching away from the interview window.
What signals does HireCredible monitor?
HireCredible monitors gaze patterns, tab switching, response timing, answer fluency, and coherence under probing follow-up questions. These signals are logged and flagged for your review — they do not automatically disqualify a candidate.
Does HireCredible use facial recognition?
No. HireCredible does not use facial recognition or emotion detection. Gaze pattern monitoring checks whether the candidate is looking toward their screen — it does not analyse facial features or emotional state.
Do integrity flags automatically reject candidates?
No. Integrity flags are surfaced for human review. You decide how to interpret them. A single tab switch may be irrelevant; multiple flags across several questions is information worth considering.
Are candidates told that integrity monitoring is active?
Yes. HireCredible's pre-interview consent screen informs candidates that the interview includes integrity monitoring. Candidates confirm consent before the interview begins.
Integrity signals built in. Zero setup required. Try HireCredible free.
Start Free →