Long-assignment integrity: using writing-development evidence responsibly
How universities can use writing-development, AI-use, reference and similarity evidence to support authorship review without treating signals as proof.
Long-form assignments create a different integrity problem from timed exams. The student may work across days or weeks, use multiple sources and, where policy permits, use AI tools. A single final document often says little about how the work developed.
Writing-development evidence can add useful context. Session history, saved versions and editing patterns can show how a submission evolved. AI-use declarations can record permitted assistance. Reference and DOI checks can identify citations that need review. Similarity analysis can highlight overlap that deserves attention.
These signals should be interpreted cautiously. A large paste may be legitimate. A sudden increase in word count may reflect work completed offline. Similarity may arise from common source material or required phrasing. A reference discrepancy may be a simple citation error.
For that reason, the evidence should lead to questions, not automatic conclusions. An authorised lecturer can review the material, consider the assessment context and, where necessary, ask the student to clarify authorship or process.
Institutions should also define which evidence is enabled, explain it to students, restrict access to authorised staff and set appropriate retention. The aim is not to reconstruct every keystroke. It is to preserve enough context to support a fair, human-led review when a concern genuinely arises.