Secure interviews

The interview you can trust the answers to.

Looking away to read. A second person in the room. A voice that is not the face. BrowseJobs interviews flag them, attach a photo to each flag, and leave the decision to your team.

Join the pilot See how it catches cheating

The problem

If the answers are not theirs, the score means nothing.

A fluent answer used to mean a prepared candidate. Now it can mean a model reading over their shoulder. Automation saves hours only if the interview at its centre cannot be faked.

How it catches cheating

Ways to cheat. Signals that catch them.

Scroll. Each pattern, and what the integrity layer reads to expose it.

01
Looking away to read Live today
Gaze drops to a phone or slides to a second screen. It is flagged as a possible, with a photo from that moment.
02
A second person in the room Live today
A second face in frame is flagged with a still, so your team can check it for themselves.
03
A voice that is not their face In pilot
In pilot: audio that leads the mouth, a looped or synthetic video, and a second voice behind the answer.
04
Evidence, not a verdict Live today
Every flag lands in the report with its time and photo. A person decides. Nothing rejects on its own.

LIVE · Integrity layerIntegrity100

GAZE · possibly looking downSECOND PERSON · possibly in frameLIP-SYNC · drift 0 ms · in pilot

Integrity report

Possibly looking down · 5:18
Possibly a second person · 7:23
Face not visible · 8:16

Photos attached. Decision: your team.

How we keep it honest

Staged demonstration. Tab, paste, face and gaze checks run in interviews today; they are camera estimates and can misfire on glasses or reflections. Lip-sync and voice analysis are in pilot. A flag sends the interview to a person and never rejects anyone on its own.

Sample report

How the report writes itself.

Scroll. The interview plays, flags land as they happen, and grading fills the scores last.

REC · 9:00 / 9:00

Monitoring · all clearPossibly looking downPossibly a second personFace not visibleCheck-in photo taken

Random check-in photo Flag

Interview report

Candidate A

Data Engineer · L1 role fit

55/100Below pass mark 60

ETL62
SQL35
Spark68
Data quality52
Data modeling45
Query optimisation60

Graded after the interview, against the role's rubric. Scores stay empty until grading finishes.

Integrity 4 flags

5:18 inPossibly looking down (phone?)
7:23 inPossibly a second person in frame
7:33 inPossibly a second person in frame
8:16 inFace not visible on camera

Check-in photos

0:46
2:16
8:00

Camera estimates. Check the photo: they can misfire on posters, reflections or glasses. Flags are noted, not punished.

Get reports like this

Sample in the format of a real BrowseJobs interview report, from a test interview recorded with consent. A bystander's face is blurred. Lip-sync and voice analysis are not in this report yet.

What is read

Four signals. One integrity score.

Gaze

Looking down or to the side, flagged as a possible with a photo. Live today.

Faces in frame

Face not visible, or a possible second person. Live today, with random check-in photos.

Lip-sync and voice

Audio against the mouth, and a second voice behind the answer. In pilot.

Answer-assist tools

Leaving the tab, the window losing focus (how an always-on-top answer tool is used), or pasting an answer. Live today.

Watch

Thirty seconds: the cheat, and the catch.

0–8s

A candidate nails every answer. Pull back: a second laptop, a phone, a friend.

8–16s

Same interview through our layer: gaze drops, a second face, a face that leaves the frame.

16–24s

Flags land in the report with a time and a photo.

24–30s

A recruiter reviews the evidence and decides. BrowseJobs: interviews you can trust.

What you receive

Evidence, not a verdict.

Every interview ends in a graded brief with an integrity section: the score, each flag, its timestamp and the clip behind it. A clean interview says so. A flagged one waits for a person.

Integrity score shown beside the rubric score, never blended into it.
A flag timeline you can scrub, with the source clip for each flag.
The candidate is told up front what is checked, and consents.
No automatic rejection. No automatic offer.

Questions before you book.

What kinds of cheating does the integrity layer look for?

Reading answers from a phone or second screen, a second person in the room, and leaving the interview to look something up are caught today from camera, tab and paste signals. A voice that does not belong to the face (lip-sync, pre-recorded or synthetic video) and a second voice are in pilot. Each flag is shown to your team with the photo it came from.

Does a flag reject the candidate?

No. A flag parks the interview for a human. Your team sees the signal, the timestamp and the clip, then decides. Automation never rejects anyone and never releases an offer.

Is it perfect?

No detector is, and we will not print a catch-rate we cannot defend. Every integrity score ships with its evidence so a false alarm is easy to overturn and a real one is easy to prove.

Do you catch answer-assist apps like Parakeet?

We catch their footprint, not their name. A browser cannot list the apps running on a computer, so we read what using one does to the interview: the window losing focus while the tab stays visible, the tab being hidden, and pasted text. Each is flagged. In practice interviews a repeat violation closes the interview. In employer interviews nothing closes on its own: the moment is noted with a photo for your team to decide. Naming and blocking specific apps would need a desktop client, which we have not built.

What does the candidate see?

They are told before the interview starts which signals are checked, and they consent to camera and microphone analysis. Nothing runs silently.

What is live today, and what is in pilot?

Live: tab and window-focus checks, paste detection, face not visible, a possible second person in frame, gaze down or to the side, and random check-in photos, all shown in the report with a photo. These are camera estimates and can misfire on glasses, posters or reflections, so every flag carries its photo. In pilot: lip-sync matching and second-voice analysis. We will say so on the report when they ship.

Onboard with us.

Choose how you want to work with us. We call you back either way.

We hire for you

Make BrowseJobs your hiring partner. You share the role on WhatsApp. Our bots screen, interview and check candidates, and keep each one updated. You meet people who already passed.

Choose this

You use our tool

Your own team runs the same bots on the roles you hire for. You stay in charge. The interview, the report and the WhatsApp updates still happen.