How interview copilots stay invisible, and what that means for honest candidates

Interview copilots put answers on the candidate's screen in windows that screen sharing does not show. A video call cannot see them. A check of the session, with consent and without live alerts, gives honest candidates a level field again.

Interview copilots listen to the question and put an answer on the candidate's screen, in a window that screen sharing does not show. Cluely advertises itself as "Invisible to screen share". Interview Coder promises that its "Overlay remains hidden when sharing your screen". For candidates who answer on their own, the interview is no longer a level field, and the people running it usually can't tell.

How do interview copilots stay invisible?

Overlay assistants draw their hints in a window that is excluded from screen capture. A candidate can share the whole screen in Zoom or Meet, and the interviewer will see a clean desktop while the answer scrolls past in the corner. The tool listens to the question and puts an answer on the screen, and the candidate reads it. On a video call, the only trace is a reading intonation and an odd rhythm in the answers, and both are easy to explain away.

The simpler versions need no special tool at all: a chatbot in voice mode on a phone next to the laptop, or a second monitor with a browser tab open. The pattern fits the tools. Copilots are best at technical questions with checkable answers, and worst at a conversation about how you handled a difficult customer.

Why can't a video call see them?

A note-taking bot or a call recorder sees audio and video. It doesn't see the window that is hidden from capture, the process running the copilot, or the second monitor the answer is on. Anything that lives inside the call is on the wrong side of the screen.

Interviewers try to compensate by watching for eyes that read and pauses that don't fit. That works sometimes, and it turns the interviewer into a detective, which is a bad use of the interview and unfair to the many honest candidates who think before they speak or look away when they do.

Who loses when cheating works?

The obvious loser is the honest candidate. Candidates for a role are compared, and every assisted answer pushes an unassisted one down the list, without anyone knowing it happened.

The employer loses later, when a hire made on assisted answers meets the actual work: the onboarding, the work redone and the second search. Gartner expects one in four candidate profiles worldwide to be fake by 2028, and in its survey of 3,000 job candidates 6% admitted to interview fraud. Both numbers point the same way.

The interviewer loses too, in a quieter way. A scorecard filled in from memory, based on answers that were partly generated, is a weak basis for a decision, and the interviewer carries it into the debrief with the same confidence as any other.

What does a fair interview look like now?

It starts with the tool that runs the interview knowing that it's an interview. A general-purpose video call can't tell. A desktop application that owns the session can check a short list of things at the level of the operating system: windows hidden from screen capture, known copilot processes, virtual cameras and audio drivers, a second monitor, and whether the person on video is live. None of this involves reading files, keystrokes outside the app or browser history. It's a check of the session, not of the person, and it only runs with the candidate's consent.

Second, a fair interview doesn't turn the interviewer into a detective. Vettasy shows no signals during the interview, on either screen. They go into a report after the session, with plain-language explanations, including the harmless ones: a streamer's virtual camera, a developer's second monitor. There is no score and no threshold. A person reads the report, and any signal can be disputed within 14 days.

Third, the same rules for everyone. When every candidate for a role gets the same plan, the same competencies and the same rubric, and the agenda is visible before the interview starts, an honest candidate knows exactly what is being assessed and has no reason to reach for a tool.

What can a candidate do?

Prefer employers who say what they check, in advance, and who are checked themselves. An employer verified by domain and legal entity, a named interviewer, an agenda before the start and a copy of the recording afterwards are not perks. They are the conditions under which answering on your own is a good strategy again.

And take the copy. If a report ever says something about your session that isn't true, the recording, the transcript and a dispute reviewed by a person are how you fix it. That is the point of symmetry: both sides see the same thing, and neither has to take the other's word for it.

Sources

• Product pages of Cluely ("Invisible to screen share") and Interview Coder ("Overlay remains hidden when sharing your screen").
• Gartner, 31 July 2025. Gartner survey shows just 26% of job applicants trust AI will fairly evaluate them. Survey of 3,000 job candidates, second quarter of 2025.
• Vettasy, Verified Session: the five signals, the report and disputes.

Written by the Vettasy team

Vettasy is a desktop application for job interviews. Employers run structured, recorded interviews in it with verified participants. It is free for candidates. Windows and macOS.