Proxy Interview Fraud: Why It's Happening More Than You Think

In 2024, a mid-size technology company quietly updated its policy for post-offer work samples. The change wasn't announced publicly. It came after three new hires — across eight months, two teams, and three different hiring managers — failed basic competency assessments within their first two weeks. One manager described the pattern plainly: "We were hiring one person and getting a different one."

They were not alone. Proxy interview fraud — where someone other than the actual candidate performs all or part of the interview — has become one of the more persistent integrity challenges in remote and AI-led hiring.

What a proxy interview actually is

The term covers a range of practices. At one end: a complete stand-in — a different person who sits in front of the camera and presents as the candidate. The candidate's name and resume are used; the stand-in is typically more technically proficient.

At the other end: a coaching intermediary. The actual candidate is on camera, but someone else is feeding answers through an earpiece, via chat on a hidden device, or by sharing control of the screen during a live coding assessment.

Between these extremes, common forms include:

• Full stand-in — a different face and voice entirely
• Earpiece coaching — real face, externally scripted answers
• Screen-share assistance — candidate answers verbally while someone else controls the coding environment
• Mid-session handoff — starts as one person, passes to another for the technical portion

Why the problem has grown

Remote interviewing removed the physical verification layer that in-person interviews provided. There is no one in the room checking identification. A laptop camera and a video call are the entire boundary between the company and the candidate.

At the same time, the economic incentive has grown considerably. A strong interview performance can be the difference between a $60,000 offer and a $140,000 one. Hiring a proxy for a one-time fee has become a cottage industry in some candidate communities — discussed openly in online forums, priced per interview or per role, with ratings and reviews.

AI tools have lowered the barrier further. Even candidates who don't hire a full stand-in can now receive real-time AI coaching that functions like one — a voice in the ear or text on a screen, answering every question faster than the candidate ever could.

What to look for

Face and voice consistency

If your process includes an early application video, phone screen, or any previous recorded interaction, compare the face and voice to the interview recording. Proxies know the resume but not the person behind it. Vocal timbre, accent, speech cadence, and mannerisms are consistent within a person and inconsistent across people — even when someone is deliberately trying to match.

Response to personal-history questions

Proxies know the resume. They don't know the memory behind it. Questions that require lived experience — "Walk me through a specific technical challenge you hit at [Company]" or "Tell me about a time your approach turned out to be wrong" — return very different answers from someone who was there versus someone who memorized bullet points. Proxies typically give either generic responses or unusually long pauses before producing something canned.

The same identity appearing across multiple candidate profiles

The hardest version to catch: the same proxy operating across multiple candidate profiles over time. If your team is running dozens or hundreds of interviews, the same person may appear under three different names, for three different roles, across three different hiring managers. No individual reviewer would see this — it requires cross-interview face matching at scale.

The cross-interview detection gap

Most integrity detection focuses on the single interview in isolation. The question being asked is: "Is something suspicious happening in this recording?" But proxy rings — where a small group of individuals interview under many names for many companies — are only visible when you look across the full dataset.

Candidate interview #31 looks perfectly clean in isolation. It's interviews #4, #17, and #31 together — three different names, the same face — that tell the real story. That pattern is invisible to any reviewer who only ever sees one interview at a time.