Proctoring Guide
What is AI proctoring?
AI proctoring uses software to observe an assessment session and identify activity that may need review. A defensible model delivers evidence to a person rather than deciding misconduct automatically. The term covers different approaches, so buyers should ask what is observed, what is missed and who makes the final decision.
Timestamped observations support a fair review.
Your institution applies its own policy and considers context.
Evidence informs a decision. A person makes it.
What is AI proctoring?
AI proctoring is a way to support assessment oversight with software-assisted observations. It may operate in a browser, review video or run on the exam device. Those approaches see different parts of a session, and the label alone does not establish what an instructor will receive when something is flagged.
For a university, the meaningful outcome is a record that can be reviewed under academic integrity policy. For an assessment platform, it is evidence that customers can use within their own workflow. Neither buyer should confuse an automated flag with proof that a student or candidate violated a rule.
How is it different from automated, live and record-and-review proctoring?
Automated proctoring records observations without a live person overseeing every moment. Live proctoring involves human oversight during the assessment. Record-and-review captures session material for a person to examine afterward. AI may assist any of these approaches, so these labels describe review timing as much as the evidence collected.
What can browser-based AI proctoring see, and what can it not?
Browser-based oversight can describe activity inside the assessment environment and support restrictions on the tab. It does not necessarily describe a separate desktop application, an invisible overlay or assistance on another device. Ask the vendor about those boundaries without assuming that a browser recording represents the entire exam environment.
What is device-level proctoring?
Device-level proctoring adds observations from the test-taker's device during the assessment window. At a high level, ScreenComply records relevant activity as timestamped evidence for review. The institution should understand the scope, disclose it to students and confirm what is not collected, rather than request unnecessary personal information.
Does AI proctoring make the decision?
It should not. The defensible model is evidence for human review, with the instructor or integrity office applying the permitted-tool policy and considering explanations. A report can identify a concern; it cannot replace the institution's process for deciding what happened, communicating the evidence and allowing a student to respond.
What does AI proctoring cost?
Common models include usage-based sessions, coverage choices and institutional or platform arrangements. Review layers, session duration, volume and custom integration requirements affect scope. Ask for a quote tied to your assessment plan rather than comparing a headline amount that may exclude the evidence, support or workflow your team needs.
What should a university ask an AI proctoring vendor?
Ask what the report contains, how a student sees the evidence and who reviews a flag first. Confirm accessibility and legitimate-tool exceptions, retention, deletion and data processing terms. Procurement should examine the student notice and review workflow as carefully as the software, because those determine whether evidence can support a fair decision.
What should an assessment platform ask?
Ask how evidence returns to your customer workflow, whether branding is configurable and how client settings and retention remain separate. Clarify the browser-based and desktop options, support ownership and pilot scope. Your customers should retain their policies and decisions even when proctoring is delivered as an integrated layer.
Frequently Asked Questions
Is AI proctoring the same as a lockdown browser?
No. A lockdown browser controls the assessment browser environment. AI proctoring is a broader term for software-assisted observation and review, which may include browser, video or device-level approaches with different coverage limits.
Can AI proctoring prove every instance of cheating?
No tool should promise complete detection. Evidence can support an informed review, but the institution must consider coverage limits, policy and legitimate explanations before reaching a decision.
Can students challenge a flag?
A fair process should let students see the relevant evidence and provide an explanation. The institution documents its review under its own policy rather than treating a software flag as a final decision.
Does proctoring need to replace our assessment platform?
Not necessarily. An integrity layer can work alongside the tools faculty and students already use. Confirm the launch and review workflow during a pilot before making a broader procurement decision.
How should we evaluate a vendor?
Ask for the report and review experience, not just a feature list. Examine what is collected, what is missed, how accommodations are handled and how retention and deletion fit institutional policy.
Evidence. Context. Human Review.
Bring your next exam to the conversation.
Review the assessment workflow, the evidence your team receives and the policy that guides a human decision.