As AI tools have become part of everyday interview prep, detecting undisclosed real-time AI use during the interview itself has become a real training focus for recruiters. One 2026 audit across nearly 20,000 AI-led interviews flagged more than a third of candidates for suspected AI assistance, and a majority of those still scored above the passing bar, which is exactly why recruiters are being trained to look closer rather than rely on the score alone. Knowing what they're trained to notice is useful even if you have nothing to hide, since a natural, well-rehearsed answer can occasionally get mistaken for a scripted one, and it helps to understand why.
Behavioral signals recruiters are trained to notice
Several behavioral patterns come up repeatedly in recruiter training. A consistent three to five second silence before every single answer, longer than natural thinking time, is one of the most cited. So is "scanning" eye movement, a left-to-right reading pattern rather than a gaze that settles naturally on the interviewer. Answers that sound like they're being read off documentation, overly polished, oddly uniform in structure, rather than recalled from memory, are another common tell, along with unusually fast or choppy typing during any written exercise. Interviewers are also trained to notice a mismatch between the polish of an answer and the natural hesitation, filler words, and small corrections that tend to show up in genuine, unscripted speech.
Technical and environment red flags
Beyond behavior, recruiters are also trained to notice technical inconsistencies: a mismatch between a candidate's stated location and their network signal, reluctance to turn on a camera unscheduled, multiple connected devices, or an unusual setup that suggests a virtual machine. Equipment or shipping addresses that don't match a candidate's claimed residence can also prompt a closer look. None of these alone proves anything, a genuine candidate can have a flaky router or a webcam that fails at the wrong moment, but recruiters are trained to weigh several signals together rather than react to a single one.
Why adaptive questioning is becoming standard
Partly in response, more interviewers now lean on adaptive questioning: changing scenarios, and follow-up "why" and "how" questions that are genuinely hard to answer well without actually understanding your own example. A scripted, generic answer that sounded fine the first time tends to fall apart under a second or third follow-up, which is precisely the point. For a candidate, the practical takeaway is the same regardless of intent: an answer you deeply understand and can defend under follow-up questions will always read as more credible than one that just sounded good once. Genuine preparation, knowing your own examples cold rather than memorizing a polished script, is what actually holds up when an interviewer starts asking why, not just what.