What can a writing-process record prove?

A strong claim is one the record can verify directly. For example, it may be able to reconstruct the captured document, check the order of recorded edits and verify server-issued checkpoints. A weaker claim moves from those facts to an inference about a person, their intent or activity the system did not observe.

The distinction matters because a technically valid record can be genuine and still be incomplete evidence for an academic-integrity decision. Verification answers, “Does this captured record pass its checks?” It does not automatically answer, “Did this student break the rules?”

The safest way to evaluate a process tool is claim by claim. Ask what the system observed directly, what its integrity checks establish and where a human inference begins.

A simple timeline can display events in order, but a verifiable record also needs a way to detect later changes. Loam links each captured edit to the previous one with SHA-256, receives server-countersigned checkpoints and replays the deltas to rebuild the final document. Its technical explanation and local verifier expose that product-specific process.

The NIST Secure Hash Standard defines SHA-256. A standard hash function alone does not validate an entire product design; the surrounding event format, key handling, checkpoint verification and final-document reconstruction still matter.

TEQSA's May 2026 guidance for higher education says students can be required to maintain a verifiable version history and discusses programs that track and replay the writing process. The same guidance says an AI score alone is insufficient for a misconduct allegation and additional evidence is required. Read the full TEQSA source for its scope and caveats.

TEQSA regulates higher education, not Australian schools, and it does not endorse Loam. Schools should apply their own assessment rules, sector requirements and procedural-fairness processes. The Australian Framework for Generative AI in Schools is the relevant national starting point for responsible and ethical use in school education.

Use the process record as one source in a documented review. The sequence below keeps technical evidence separate from the final academic decision:

1. 01

   State the rule

   Identify the task conditions and the assistance that was permitted, required or prohibited.

2. 02

   Check the record

   Confirm what was captured, whether integrity checks pass and where the record has gaps.

3. 03

   Read in context

   Review the final work and inspect only relevant moments rather than treating every unusual pattern as suspicious.

4. 04

   Hear the student

   Ask for an explanation and consider accessibility needs, technical problems and legitimate working methods.

5. 05

   Make and record the decision

   Apply school policy to all relevant evidence, document the reasoning and preserve the ordinary review or appeal path.

Ask questions that force the boundary into the open. The answers should be specific enough for teaching, privacy and IT reviewers to test:

• Which actions are captured, and which are outside the record?
• Can later changes, deletions or reordered events be detected?
• Who controls the evidence, access and retention period?
• What student data, device data or behavioural signals are collected?
• Can a reviewer inspect a synthetic sample before procurement?
• What happens when connectivity fails or a student needs an accommodation?
• Can an exported record be verified without trusting a screenshot?
• Does the product issue a verdict, or preserve evidence for human review?

For Loam specifically, start with the public verifier and technical boundary, then review the Privacy Policy and Data Processing Agreement.