AI and Academic Integrity

1. Our position

Graspera uses AI to generate practice questions, mark answers and explain solutions. We think that is a good use of the technology, and we also think students and schools are owed a straight account of what it does, what it cannot do, and where the line sits between practising with it and cheating with it.

This statement forms part of our Terms of Service.

2. How the AI is used

The models are supplied by Anthropic PBC. The Privacy Policy explains exactly what is sent to them, what is not, and that no one trains a model on student work.

3. What we do to keep it accurate

A general-purpose language model on its own is not reliable enough to mark schoolwork. We do not use one on its own.

1. Questions are generated from templates built against the syllabus, so the content stays inside the module and at the right level, rather than being invented freely.
2. For subjects with calculable answers, an independent solver written in code works the problem out separately. Where the solver and the AI disagree, the question is not shown to the student.
3. Marking is done against an explicit marking scheme rather than an open-ended judgement.
4. Generated content is reviewed against the published syllabus and, where relevant, against past examination marking guidelines.
5. We record which syllabus version content was authored against, so material can be found and revised when a syllabus changes.

These checks reduce errors. They do not eliminate them.

4. Where the AI can be expected to fail

We would rather name these than have a student discover them at the wrong moment.

• It can mark a correct answer as wrong, particularly where the student used a valid but unusual method, or wrote the answer in a form the marking scheme did not anticipate.
• It can mark a wrong answer as correct, particularly in extended written responses where the reasoning matters more than the conclusion.
• It can state something confidently and be wrong. Confidence is not a signal of accuracy.
• It may not reflect the very latest syllabus change or NESA advice.
• It handles diagrams, graphs and hand-drawn working far less reliably than typed text.
• It is weaker on extended-response marking - the kind of question where a human marker weighs an argument - than on questions with a determinate answer.

5. Academic integrity

5.1 The rule

5.2 Fine to do

• Practise questions and check your answers.
• Ask why a method works, or why your approach lost marks.
• Ask for an explanation in different words when the textbook is not landing.
• Ask for more questions on the thing you keep getting wrong.
• Use the feedback to redraft your own work in your own words, then check it again.

5.3 Not fine

• Copying a Graspera answer into an assignment, report or take-home task.
• Paraphrasing a Graspera answer and handing it in as your own thinking.
• Using Graspera during an assessment your school requires you to complete unaided.
• Using it to write a depth study, a research task or a practical report for you.

5.4 NESA and your school

HSC students complete an all-my-own-work requirement and are bound by the NSW Education Standards Authority rules on malpractice, as well as by their own school's assessment policy. Those rules govern, and they are stricter than anything we could write here. Breaching them can mean zero marks for a task, and in serious cases can affect an HSC result.

If you are unsure whether using Graspera for a particular task is allowed, ask your teacher before you use it, not after.

5.5 For schools

We support schools that want to set clear rules for how Graspera may be used. We do not offer, and will not offer, a feature designed to produce work for submission. If your school assessment policy needs a written statement from us for its own records, email support@graspera.au and we will provide one.

6. The principles we hold ourselves to

• The AI does not replace the teacher. It is a way to get more practice and faster feedback between lessons, not a substitute for teaching.
• A student is never marked in secret. A parent can see what a student was asked, what they answered, and how it was marked.
• No automated decision with real consequences. Graspera does not stream, rank, grade or gatekeep a student, and does not send a score anywhere that affects their standing.
• Errors are surfaced, not hidden. Where our own checks disagree with the AI, we withhold the question rather than show something we cannot stand behind.
• Student work is not a training resource. Not for us, not for our provider.
• No engagement mechanics aimed at children. No streak pressure designed to make a child anxious about stopping, no advertising, no notifications designed to pull a student back at the expense of sleep or family time.

7. Telling us when it gets something wrong

Report errors to support@graspera.au. Tell us the subject and module, what the question was, what you answered, and what the AI said. A screenshot is ideal.

We treat a wrongly marked answer as a defect, not a complaint. Reports of a safety problem or an inappropriate response are actioned first.

8. Changes

AI capability, and the rules around its use in schools, are both moving quickly. We will update this statement as they do, and material changes are notified to Account Holders. The version and date at the top of this page always show the current version.