Review queue
Observations under human review.
No corrected observation becomes employer-visible until the candidate has reviewed the outcome.
Attempted personality categorization
Fictional candidate · Detected July 10, 2026
AI output described a candidate as 'introverted overall' without evidence. Blocked from the employer surface pending human review.
Proposed correction
Remove the personality label. Retain the evidence-tied observation about listening carefully in group settings.
While a review is pending, employers see “Under Further Review” in place of the original observation. The candidate’s private written context is not shown here.
Confidence label mismatch
Fictional candidate · Detected July 9, 2026
A high-confidence claim was based on a single 12-second response. Proposed reduction to moderate before publication.
Proposed correction
Reduce confidence from high to moderate and add a note about limited supporting evidence.
While a review is pending, employers see “Under Further Review” in place of the original observation. The candidate’s private written context is not shown here.