Virtual interviews are a standard part of the hiring process because they’re faster and more convenient for candidates and organizations. But they’re also vulnerable to new forms of fraud that candidates use to land jobs they aren’t qualified for.
Instead of candidates completing the interview on their own, they use proxy interviews, AI interview help tools and assistants, and even deepfake AI (we promise this isn’t just a sci-fi buzzword, and it’s particularly concerning for phone interviews).
These are used more often than you may realize, especially proxy interviews and AI interview helpers, because they’re good enough to help them fake it ‘til they make it into your organization.
We’ll show you how these work and which tools your organization needs to detect and block AI fraud during traditional virtual interviews and online technical interviews.
How does interview fraud actually happen?
Proxy interviews
What is a proxy interview?
A proxy interview is when someone other than the candidate (a proxy) completes an interview or pre-hire assessment for the candidate. An interview proxy can be a paid subject matter expert or just a friend or colleague who can help the candidate pass.
How do proxy interviews work?
Candidates share their login credentials with the proxy, which gives them access to the virtual interview platform, such as Zoom or GoToMeeting, or remote access software like TeamViewer and AnyDesk, which allows the proxy to control the candidate’s device.
Here’s how proxy interviews work for video and technical interviews:
- During traditional spoken virtual interviews, the proxy is either on camera completing the interview as the candidate, or the candidate stays on-screen while the proxy feeds them responses.
- On the other hand, the candidate usually stays on camera during technical interviews and pre-hire skills assessments that don’t require spoken responses because the proxy controls their device and does everything—solves the problems, writes code, navigates the screen—while the candidate is on the screen pretending they’re doing the work.
AI interview help tools, copilots, and assistants
Candidates aren’t just using AI to write their resumes; they’re also using AI interview help tools, which are also called interview copilots and assistants, such as Cluely and FinalRound. These tools are becoming more common because they’re cheap, hard to detect, and smart enough to help candidates land jobs they aren’t qualified for in basically any industry.
How do AI interview helpers work?
AI interview helpers are basically chatbot teleprompters that listen to the interviewer’s questions and feed responses quickly enough for the candidate to read them back and still sound natural. If there’s a brief pause before they read, candidates can simply get ahead of it with a “sorry if my audio is delayed, my internet’s been slow today” excuse.
These interview assistants can also read what’s on the candidate’s screen, like tasks inside a development platform or questions in a pre-hire assessment. That means a candidate can use a coding-specific assistant, and the AI can write the code, provide step-by-step instructions, and answer related questions during the assessment.
This type of interview AI is also called an “invisible AI assistant” because it’s REALLY difficult to detect because it:
- Sits on screen as a translucent overlay that’s even invisible during screen sharing.
- Works on a cell phone and other devices placed just outside of the webcam’s view, listens to interview audio, and gives real-time responses.
- Uses hidden keyboard commands to reposition the overlay and ask the AI follow-up questions without obvious on-screen activity.
Deepfake AI use during interviews
Deepfakes used to look like poorly dubbed movies with mismatched lips and avatars that never blink, but new deepfake AI models blink, move their lips naturally, and some even adjust to different lighting, as you’ll see in the video below.
What are deepfakes?
Deepfakes use AI to manipulate a person’s face, voice, and gestures in real-time during live video and phone calls. The AI is trained using photos, videos, and audio recordings to create a fake version of the person’s face.
How can deepfake AI be used in job interviews?
When deepfake AI is trained properly, it can look and sound like the real candidate, which is even more difficult to detect if the interviewer hasn’t met or spoken with the candidate before. It can be used during:
- Video interviews: The “candidate” you see on video may just be a face-swap or AI avatar lip-syncing in real-time.
- Phone interviews: Cloned voices mask the real speaker’s voice in real-time during the conversation.
The video above contains clips from a longer video by creator, r4f43lo. It was shortened and formatted for the purpose of this article.
How to prevent interview fraud
Verify identity
Confirming who’s completing the interview is a basic but important part of interview security because they may end up working at your organization and potentially have access to your organization’s data and proprietary information.
This step needs to be secure, but it also needs to be quick and simple so that it doesn’t distract the interviewee or add unnecessary stress. As this involves sensitive personal data, ask the service how they store and protect data and how long they keep it.
Allow required platforms, block everything else
Organizations need the ability to allow specific applications and platforms required for the interview while blocking AI chatbots, remote access tools, AI coding tools and interview assistants, screen recorders, and internet searches.
For example, restricting access to all applications, platforms, and browsers except for those needed to complete the interview, such as:
- Web conferencing tools like Zoom and Google Meet, for virtual job interviews
- Job-specific platforms and tools like AWS, Azure, Excel, and Word for technical interviews that require candidates to show their competency with tools they’ll use on the job.
The only realistic way to do this is to use a hybrid proctoring solution because locking the browser isn’t enough. But a hybrid solution lets you provide access to the exact platforms, applications, and sites candidates can use while blocking everything else.
Detect cell phones, smartwatches, and other devices
A job candidate wearing a watch to an in-person interview is normal, but it’s a concern during a virtual interview, especially technical interviews, because smartwatches can give them access to notes or AI, just like cell phones and tablets can.
You have two options to catch candidates using cell phones and other devices: hoping the interviewer or a live proctor notices one of them, or using multi-device detection technology that detects cell phones and other secondary devices and, for technical interviews, flags phrases like “Hey Siri” and “Alexa” that activate virtual assistants on those devices for help.
Record and monitor behavior and the environment
The candidate’s desktop can be recorded, and their workspace can be recorded and monitored using their webcam, with an optional second camera for a complete view. Interviewers should be able to access these recordings in HD with timestamped flags, side-by-side video and screen playback.
Protect your organization’s IP
In addition to stopping interview fraud, your organization also needs to protect its intellectual property by securing the questions and content used in technical interviews and pre-hire assessments. Search & DestroyTM automatically scans the internet for your organization’s leaked content in minutes and lets you send one-click takedown requests.
Online proctoring can stop interview fraud
Whether it’s a virtual interview over Zoom, a technical interview that involves real job tasks, or a pre-hire exam, interview fraud puts your hiring decisions, teams, reputation and credibility, AND data at risk.
Honorlock’s interview proctoring detects and stops interview fraud so organizations can interview, assess, and hire candidates with confidence at scale.
