Why every proctoring flag needs a person
A proctoring flag is a signal that tells a reviewer where to look. It can’t tell them what happened. A candidate looks away to think, a sibling walks into the room, the lighting changes, a connection drops, or a student with a disability moves in a way the model wasn’t expecting. Any of these can raise a flag, and none of them is misconduct.
These are proctoring false positives, and every proctoring product produces some. What matters for an institution is what happens next. When flags go straight to a penalty, honest candidates pay for the limits of the technology. When a person reviews each flag in context, records a reason and passes only confirmed incidents to a committee, the process is fair and the record holds up. For what the AI looks for in the first place, read our guide to AI proctoring.
How a flag becomes a reviewed case
- The AI raises a flag with the clip, snapshots and events around it.
- A reviewer opens the case at the flagged moment and looks at it in context.
- The reviewer records a decision with a required reason: no incident, or a possible incident to escalate.
- Your institution decides. Confirmed incidents go to your malpractice workflow, where a committee records the final decision.
The reviewer console
Each case opens in the reviewer console at the flagged moment, with webcam, screen and events on one timeline. Reviewers make one-key decisions and have to give a reason, so no flag is dismissed or escalated without an explanation. Escalating a case takes one step.
The review priority score
Each candidate gets a review priority score that shows the signals behind it. The score is labeled a priority for review, never a finding, and it helps reviewers choose which cases to open first during a busy sitting. It doesn’t say a candidate cheated, and it doesn’t score emotion, personality or “suspicion”.
Every flag reviewed by Proctorly
At the Every flag reviewed level, a trained Proctorly reviewer looks at each automated flag before it reaches you, so only reviewed cases land in your queue. Reviewers work 24×7 during exams, in English, with one reviewer for every 40 candidates. They review 95 of every 100 flags within 5 minutes, and the rest are still reviewed, just later. Institutions can set their own turnaround.
“Every flag reviewed” covers every flag. It doesn’t mean someone watches every minute of every recording. Full-session review as a separate option is planned.
Record and review with your own staff
At the Recorded level, Proctorly records the session and raises flags, and your own trained staff review cases in the same reviewer console. Record and review proctoring suits quizzes and internal assessments, where faculty know the candidates and the stakes are lower. If you’d like our people to handle review for higher-stakes exams, record and review proctoring services are part of live proctoring and our other exam services.
Proctoring evidence reports that hold up
Every disputed result needs a record, because boards and appeal panels will ask what happened, what was seen and who decided. Proctorly keeps that record for every candidate.
- Case record. One record per candidate, listing each identity check, flag, nudge, clip and reviewer decision in order.
- Reasons as well as outcomes. Each reviewer decision carries the reason the reviewer gave.
- Tamper-evident audit trail. Every view, decision and change is logged, so you can show the record hasn’t been altered.
- Evidence pack. One export per case with clips, timeline, decisions and audit entries, for an appeal panel, a regulator or an internal audit.
Identity checks go into the same record. See how identity verification for online exams is logged, check by check.
From confirmed incident to committee decision
Even a confirmed incident doesn’t trigger a penalty. On Proctorly, the malpractice workflow sends confirmed incidents to a recorded committee decision, and no penalty applies before it. Your institution names who sits on the committee and how candidates appeal.
In online exam appeals, the evidence pack gives the committee and the candidate the same facts: what was flagged, what the reviewer saw, the reason they recorded and who decided. If you run the Proctorly Examination module, hall and online incidents go through one malpractice workflow.
The exam-day control room
Your exam office can see every online sitting live, including who has started, who has been flagged and who needs help.
How the process limits proctoring false positives
Human review is the main safeguard, and the rest of the process is built around it.
- Signals, never findings. Unusual audio and movement are raised for a person to review in context. Reading aloud, assistive tools, household noise and disability-related movement are not misconduct.
- Reminders before flags. Before anything is flagged, on-screen nudges remind candidates of rules such as staying in view. Every nudge is logged.
- Required reasons. A reviewer can’t record a decision without a reason, which makes inconsistent calls easy to spot.
- Fairness monitoring. Detection is checked for fairness across devices, networks, lighting and accommodations.
- Technical gaps kept apart. Five evidence labels are planned to keep a camera or upload failure in its own category, outside the integrity score.
- Your rules for pausing or ending a session. We plan to make automatic pause or end of a session a setting your institution controls, switched off unless you turn it on. An ended session is never treated as a finding of misconduct.
What happens if you get flagged on a proctored exam
This section is for candidates. If you’re reading it because you were flagged, this is what a flag means on Proctorly.
Being flagged doesn’t mean you have been found to have cheated. It means the software noticed something a person should look at. A trained reviewer watches the clip and records a decision with a reason. If the reviewer thinks there may have been an incident, the case goes to your institution, and only your institution decides the outcome, through its own committee and appeal process.
If you were flagged for cheating on a proctored exam and think it was a mistake, ask your institution about its appeal process and what evidence was recorded. If something went wrong during the exam, such as a dropped connection, tell your institution and our support team as soon as you can.
You may also have searched for terms like “honorlock flagged”, “proctorio flagged” or “flagged on proctorio”. Each proctoring vendor, Honorlock and Proctorio included, handles flags in its own way, and your institution decides how flags lead to outcomes. Check the vendor’s current documentation and your institution’s policy to see how a flag on that system is reviewed.
Exam integrity matters to everyone who sits honestly, and flags exist so a fair, recorded review can take place.
Honest about what we see
No proctoring product sees everything, and a reviewer can only judge what was recorded. Our commitments are already public: an AI flag is never a finding, monitoring is never hidden, and we never claim to catch every form of cheating. For each setup we publish a summary of what we can and cannot see.
Evidence labels, notices and review controls
We plan to add the following to the review process. None of them is available yet.
- Five evidence labels. Observed, Suspected, Blocked, Unobservable and Unavailable, so a technical failure stays in its own category, outside the integrity score.
- A notice to the candidate when a case is opened, plus a reconsideration step.
- A reviewer role that sees only assigned exams and can’t change marks, change your policies or impose a sanction.
- Institution-controlled rules for pausing or ending a session.
- Blind double review to check that reviewers are consistent.
- An AI case summary that helps the reviewer and never decides.
Frequently asked questions
What is human review proctoring?
It means a trained person looks at every automated flag before anything happens to the candidate. On Proctorly, the reviewer opens the case at the flagged moment, sees webcam, screen and events on one timeline and records a decision with a reason. Only confirmed incidents reach your institution, and your committee makes the final decision. The AI raises flags and never decides an outcome.
Who reviews the flags: our staff or Proctorly?
You choose for each exam. At the Recorded level, your own trained staff review cases in the reviewer console. At Every flag reviewed and the live levels, a trained Proctorly reviewer looks at each flag first. Our reviewers work 24×7 during exams in English, with one reviewer for every 40 candidates. In both cases your institution names who makes the final decision and how candidates appeal.
How quickly are flags reviewed?
At the Every flag reviewed level, Proctorly reviewers review 95 of every 100 flags within 5 minutes, and the remaining flags are still reviewed, just later. Institutions can set their own turnaround to fit their results calendar. The level covers every flag, which is different from watching every minute of every recording. Full-session review as a separate option is planned.
What does an evidence pack contain for an appeal?
An evidence pack is one export per case. It holds the relevant clips, the full timeline of identity checks, flags and nudges, each reviewer’s decision and reason, and the tamper-evident audit trail entries showing who viewed or changed what. An appeal panel, regulator or auditor gets the same record your committee used and can trace the decision step by step.
How do you reduce proctoring false positives?
Every flag goes to a person who sees it in context, and every decision needs a reason. Audio and movement count only as signals, and reading aloud, assistive tools or disability-related movement are not misconduct. Candidates get logged reminders before any flag, and detection is checked for fairness across devices, networks and lighting. Evidence labels that keep technical failures out of the integrity score are planned.