Cheating is now the default, not the exception
Across 19,368 technical interviews, 38.5% of candidates were flagged for AI-assisted cheating, and the rate climbed from 9% to 45% in a single quarter.
Fabric · State of AI Interview CheatingAn agent runs on the candidate’s machine during the interview and flags AI assistants, hidden overlays, remote control and virtual devices as they appear. No screen or audio recording. Every session ends in a signed, tamper-evident report.
In under a year, AI assistance during live interviews went from a fringe trick to a funded, mainstream product category. If your interviews aren’t monitored, you may already be extending offers on answers that were never the candidate’s.
Across 19,368 technical interviews, 38.5% of candidates were flagged for AI-assisted cheating, and the rate climbed from 9% to 45% in a single quarter.
Fabric · State of AI Interview CheatingCluely, the “cheat on everything” overlay that began as Interview Coder, raised $15M from Andreessen Horowitz. Undetectable interview assistants are now a funded, fast-growing market.
TechCrunch · Cluely raises $15M from a16zGartner predicts that by 2028, 1 in 4 candidate profiles worldwide will be fake, driven by AI-generated identities and stand-in, proxy interviewees.
Gartner, via HR DiveInterviewWatch isn't a proctoring tool. It runs a consent-first agent for the length of the interview, watching only machine-level signals, AI overlays, extra displays, clipboard, remote access, virtual cameras, and turns them into a live risk score and a signed report. No screen, audio, or video, ever.
The candidate consents, the agent runs only for the session and exits when it ends, no personal data captured, no media captured.
Observe integrity metadata during the session, detect high-risk patterns and surface issues while the interview is still happening.
Give recruiters, hiring managers and compliance reviewers the same signed record of what happened and why it matters.
InterviewWatch detects signals of AI-assisted behavior during the interview and surfaces them in real time. Every detection is timestamped as it happens, so interviewers can see potential assistance immediately instead of discovering it after the interview is over.
Review the complete timeline after the interview. See what was detected, when it happened, how long it lasted, and how it affected the integrity score, giving your team a clear record of the session.
The candidate opted in and monitoring started cleanly at 00:04. Nothing is recorded before consent is given, so the session begins with a verifiable clean baseline.
ChatGPT was opened in a foreground window at 06:12, visible on the candidate's own screen. An AI assistant being open during the interview is treated as high severity, whether or not it was used.
Active window focus moved to Claude for 9 seconds at 07:40, meaning the candidate briefly brought an AI assistant to the front during the interview.
More than one monitor was connected to the candidate's machine during the interview. A second screen can display reference material that isn't visible to the meeting camera.
AnyDesk was running on the candidate's machine from 10:15 to 10:16, 63 seconds total. This software is capable of letting another person view or control the machine remotely.
ChatGPT Desktop was running with a window deliberately excluded from screen capture from 14:02 to 14:19, 17 seconds total. Overlays of this kind are visible to the candidate but invisible to the interviewer's screen-share view.
A 428-character copy was taken from ChatGPT at 14:18, then available to paste into the interview. We record the source app, window title, and length, never the clipboard contents.
This is a representative example built from InterviewWatch's real signal catalog. Every actual report is unique to that interview and its own evidence.
InterviewWatch draws a bright line between integrity metadata and interview content, so teams can protect hiring quality while keeping candidate privacy clear. The agent collects no personal data, it only watches for machine-level integrity signals.
Your files, code and work stay untouched.
Nothing is captured or stored as media.
No access to your file system or browsing.
No keystroke logging, no clipboard access.
That is the setup, and it is the same whether interviews come from Outlook, Google Calendar, an ATS or a scheduling tool. Nothing to install, no API key, no integration to build.
Optional, so it never reads as someone expected to speak. We reply on the thread within seconds with your session link, and a reschedule moves the session with it.
They get their own email with the download and a plain list of what is and is not observed. Nothing is collected until they accept. Windows 10 and 11.
Nothing to start, nothing to stop, and no timer to set. The signed report comes back on the same thread once the interview is over.
A monthly fee for the account, then a charge for each interview that is monitored. A meeting nobody joined is not billed. No minimum, no contract.
Five monitored interviews, to read a real signed report before paying anything.
Plus $5 per interview, charged on completion. An interview nobody joined is never billed.
For high-volume teams that need governance, custom terms and a security review.
Straight answers for legal, security, recruiting operations and hiring managers.
In 20 minutes, we will walk through consent-first live interview monitoring, report review and how the hosted platform fits your team.