This Is Happening More Than You Think
In 2024, a software company in Austin made four engineering hires in three months. By month six, all four had been let go. None of them could actually do the work they'd demonstrated in interviews. An internal investigation found that at least two had used a paid third party to complete their technical screens.
Proxy hiring and deepfake interviews are now a real operational risk for any company doing remote hiring, and most companies aren't adequately prepared for it.
What Proxy Hiring Actually Is
Proxy hiring means someone other than the actual candidate is completing part or all of the hiring process. This ranges from a friend helping with a take-home test, to a paid professional completing an entire interview on someone else's behalf via screen share.
The paid proxy industry has become surprisingly organized. Forums on Reddit and Telegram openly advertise services for remote coding interviews. Prices range from $200 for a technical screen to $2,000 for a full-day onsite simulation.
How Deepfake Video and AI Voice Are Being Used
Deepfake technology has crossed a threshold where real-time video manipulation is possible on consumer hardware. Tools like DeepFaceLive can overlay a different face onto a live video feed. AI voice cloning can replicate someone else's voice with a few minutes of audio training data.
In practice, this means a candidate can appear on a video interview as themselves, but have a third party feeding them answers through an earpiece while AI overlays maintain a convincing visual.
How to Spot It: Behavioral Signals
The most reliable signals are behavioral, not technical. Here's what to watch for during live interviews.
- Unusual latency in responses. A pause of 2-4 seconds before every answer can indicate the candidate is receiving information via earpiece or reading from another screen.
- Eye movement patterns. Eyes that consistently track to the same spot off-camera, especially moving in a reading pattern rather than natural thinking patterns, are a red flag.
- Inconsistency between interview and follow-up. Ask the candidate to explain something from their test in a follow-up call. If they can't, that's telling.
- Overly rehearsed language. Answers that sound scripted, with no "ums", no self-corrections, and perfect structure, can indicate read-aloud responses.
What You Can Do Today
You don't need to wait for your platform to add detection features. There are practical steps you can take right now.
- Require ID verification at the start of every remote interview and match it to any previous ID on file.
- Include at least one unscripted, contextual question that references specific details from the candidate's resume. A proxy won't know those details.
- Follow up every technical screen with a short verbal explanation session. Ask the candidate to walk you through their logic.
- Pay attention to the gap between written/recorded performance and live, conversational performance. That gap is the clearest signal.
The goal isn't to assume every candidate is cheating. It's to build a process where cheating is hard enough that honest candidates aren't disadvantaged by it.