What is AI Assistance Detection?
AI Assistance Detection is a premium feature that provides recruiters with additional signals indicating when an interview response may show patterns consistent with the use of AI tools during an interview.
The feature is designed to support informed review and decision-making. It does not make hiring decisions or take automated action on candidates.
Is AI Assistance Detection included by default?
No. AI Assistance Detection is a premium feature.
To request access, please contact your Customer Success Manager, who can confirm availability and help enable the feature for your account.
How is AI Assistance Detection configured?
AI Assistance Detection is configurable at the job level.
Once enabled for your account:
- Recruiters can choose whether to enable the feature on individual jobs
- This allows teams to apply the feature selectively based on role requirements or interview type
How does AI Assistance Detection work?
When enabled for a job, AIR analyzes completed interviews using an AI proctoring model.
The model reviews:
- Interview transcripts
- Response timing and sequencing data
Rather than relying on any single response, the model looks for patterns across the full interview that may be consistent with AI-assisted or scripted responses. These patterns can include combinations of linguistic consistency, response structure, and timing behavior that differ from typical spontaneous, experience-based answers.
Based on this analysis, AIR generates a likelihood score indicating whether AI assistance may have been used during the interview.
To reduce the risk of incorrect flags, AIR is intentionally conservative and surfaces results only when confidence is high.
When is an interview flagged?
An interview is flagged only when the model identifies a high likelihood of AI assistance.
When this threshold is met:
- “AI assistance detected” appears on the interview table
- This label is clickable, allowing recruiters to view more details
The same information is also visible on the interview video page.
When reviewing a flagged interview, recruiters can view:
- The confidence score
- A high-level explanation of why the interview was flagged
- Relevant context alongside the interview video and transcript
This information is intended to support review and interpretation, not to make automated judgments.
Does this mean a candidate definitely used AI?
No.
AI Assistance Detection does not determine intent, confirm behavior, or label a candidate as having violated any policy. It provides a probabilistic signal, not a definitive judgment.
We recommend that recruiters:
- Review flagged interviews in context
- Consider role expectations and interview format
- Use this signal alongside other evaluation criteria
Does AIR take automated action on candidates?
No.
AIR does not:
- Disqualify candidates
- Block interview progression
- Notify candidates
- Enforce hiring decisions
All review and decision-making remains fully under recruiter control.
How does AIR reduce the risk of incorrect flags?
The detection model is intentionally conservative and designed to prioritize precision over coverage.
AIR only surfaces signals when confidence is high, helping reduce the risk of mislabeling candidates and supporting responsible use in hiring workflows.
Is AI Assistance Detection meant to replace recruiter judgment?
No.
AI Assistance Detection is a decision-support feature, not a substitute for human review. It is intended to provide additional context so recruiters can make fair, informed decisions.