Integrity
Integrity isn’t guesswork.
Screen watches for the signals that matter in a remote technical interview and flags them with evidence. A person reviews every flag. Nothing is auto-rejected.
What Screen looks for
Signals, not suspicion.
Each signal is logged with a timestamp and detail, so a reviewer can see exactly what happened and when.
Visible proctoring
The candidate can see the camera is on. Screen notes when no face is present, attention leaves the screen, an object appears in frame, the tab changes, fullscreen exits or screen-share drops.
Overlays & assistance tools
Flags behaviour consistent with on-screen overlay assistants — such as reading-pattern eye movement or answers delivered as if read aloud. Inferred from signals, not detected as specific apps.
AI-assisted answers
Where enabled for a role, answers are checked for patterns consistent with generated text or a sudden jump in capability.
Paste anomalies
Large, pre-formed blocks pasted into the code editor are flagged with their size and timing.
Latency anomalies
Response-timing patterns consistent with looking answers up elsewhere are flagged for review.
Multiple faces
Another person appearing in frame is recorded with when and for how long.
A second voice
After the interview, the recording is checked for a second or whispering voice. Findings go to review only.
Combined signals
Weak signals that occur together are escalated as one finding, so reviewers see the pattern.
Principles
Flag, show, let a person decide.
Every flag goes to human review
Flags are inputs for your reviewers, not verdicts.
Never auto-rejected
No candidate is rejected automatically because of an integrity flag.
Scored separately
Integrity never moves the competency score. Reviewers see both, side by side.
Evidence attached
Each flag carries its timestamp, duration and detail, so decisions can be explained.
No system can detect every form of assistance. Screen’s job is to surface evidence so your reviewers can make a fair, informed call.