AI Detection reports are embedded workbooks. Their tabs, filters, and summary metrics can vary by
workspace and report version. Follow the labels and metric descriptions in your report; the
examples below may not all appear.
Understand the result
AI classification
AI classification
The report summarizes the processed submission as AI Only, Mixed, or Human Only.
This label is a model output, not a finding of misconduct.
AI confidence or probability
AI confidence or probability
The overall percentage expresses the detector’s assessment of the submission as a whole. It is
not the percentage of words written by AI and should not become an automatic decision threshold.
Sentence detection
Sentence detection
Highlighting identifies sentences that the detector marked for closer review. Each sentence is
evaluated separately from the overall submission result.
Review the submission
1
Verify the record
Confirm the student, course, assignment, submission type, and submission date.
2
Read the complete work
Read the submission before you focus on highlighting. Note quotations, prompts, templates,
citations, and other structured text.
3
Compare the signals
Consider the classification, overall score, and highlighted sentences together. Expect them to
reflect different levels of analysis.
4
Add assignment context
Check permitted tools, disclosure requirements, drafts, prior writing, and relevant instructor
observations.
5
Follow institutional policy
Use the required human review, documentation, communication, and appeal process.
Explain apparent mismatches
The overall score and sentence highlighting do not need to move together:- A low overall result can have no highlighted sentences.
- Many highlighted sentences can appear even when the overall result is below 100%.
- A Mixed classification can contain both highlighted and unhighlighted text.