Stories about proxy interviewees, borrowed identities and AI-assisted answers have made hiring teams nervous. If you hire remotely, particularly in cyber and AI security, interview fraud is a risk you should plan for. It is also a topic where a lot of vendors overpromise.
This article explains what a good integrity report can honestly tell you, what it cannot, and how to follow up fairly.
The three problems people mean by "fraud"
- Impersonation: someone else attends the interview or the person is not who they claim to be.
- Assisted answers: the candidate is reading from a script, another person, or an AI tool.
- Misrepresentation: the CV claims things the candidate cannot discuss.
Each needs a different response, and none can be detected with certainty from a video call.
What a credible integrity report does
A trustworthy report is modest. In Critical Hive the integrity report lists five checks:
- Identity capture: was a selfie provided at the start, and can the recruiter see it next to the report?
- Live presence: was the camera on during the interview? A few still frames are checked and discarded; video is not stored.
- Answer authenticity: signals such as pasted text or bursts of typing that are worth a question.
- Attention: did the candidate stay in the interview window?
- Claim consistency: where a CV was supplied, do answers match it?
It then gives an assurance level (strong, partial or low) and, importantly, says what was not checked.
What it cannot do
- It does not prove anyone cheated. A flag is a question, not a verdict.
- It does not match the selfie to a government ID.
- It does not detect deepfakes with certainty, and we do not claim to.
- It does not read minds: a candidate may glance away because a child walked in.
If a vendor claims it can detect AI-written text with certainty, ask for independent evidence. Such detectors are widely reported to produce false positives, which is a serious problem when a person's livelihood is at stake.
A fair follow-up process
- Never reject on a flag alone.
- Use the questions the report suggests in a short live video call.
- Ask the candidate to walk through a recent project in detail: real experience holds up under follow-up questions.
- For roles that need it, verify identity and credentials in person or through an accredited provider.
- Record what you decided and why.
Design the process so fraud is harder
- Use scenario questions that need reasoning, not recall.
- Ask for specifics: what exactly did you do, what happened as a result?
- Keep a human interview early for senior or sensitive roles.
- Verify before access, not after.
For cyber and AI-security roles
These roles carry elevated risk because a fake hire can gain access to sensitive systems. Combine structured interviewing with reference checks and, for cleared roles, the clearance process. Our guide to hiring AI security specialists covers what to test.
FAQ
Can AI detect fake candidates?
It can surface signals that deserve a human look. It cannot prove fraud, and you should not treat it as a verdict.
Is it legal to use a selfie in the interview?
Collecting identity images raises privacy duties. Be transparent, collect only what you need, restrict who can see it and set a retention period. See AI interviews and the Privacy Act.
Does Critical Hive record video?
No. Video is not recorded or stored.
See an example
Open the live demo to look at a real integrity report, or start a pilot.