Using AI in Leibniz

Leibniz does not offer AI chat. Students practise from a question bank. In the background, AI can add questions and worked solutions for syllabus topics when the bank needs more content.

AI-assisted marking is separate from question generation. It compares the working you submit with the question and marking context, then returns a mark and feedback. AI models generate responses from patterns learned during training, so they can make mistakes even when an explanation sounds confident.

Class comparisons and practice statistics are calculated from recorded results. They are not AI-generated predictions.

Try the question yourself, show your steps clearly and use the solution to understand the method. Check calculations, diagrams and assumptions against your course materials. If feedback seems wrong, ask your teacher or report it rather than assuming your work is incorrect.

AI can misread handwriting, miss valid alternative methods or produce biased, unfair or inappropriate content. Teachers should check suitability for their students and review disputed feedback. AI practice feedback should not be the sole basis for a formal grade or a decision about a student.

Follow your school’s rules and applicable laws on permitted AI assistance, attribution and copyright. Do not submit generated solutions as your own work where this is prohibited. Only upload material you are authorised to use or share.

Submit only the answer and working needed for the task. Crop out names, student numbers and unrelated information from handwritten work. Never include passwords, private messages or information about someone else.

AI feedback requires processing the working you submit through AI providers. Before submitting, read our privacy policy; schools can also read the data processing agreement. Ask your teacher or contact us if you are unsure whether you have permission to share material or want to discuss privacy choices. Obtain any consent required for the particular use: opening these guidance links does not record consent.

Use the question’s report control where available, or email letterbox@leibniz.com.au for incorrect answers, unfair feedback, unsafe content or privacy concerns. Include the question reference and a short description; avoid adding other students’ information. Ask your teacher for help if needed.

We review reports and can investigate or correct content. For disputed practice feedback, request a human review rather than relying on an AI response alone.

For a planned relocation or expansion of AI processing into another country, we provide at least seven calendar days’ advance written notice to affected schools and direct account holders. The notice identifies the affected feature, provider, countries, data involved and effective date. School notices are provided to the nominated administrator for sharing with affected users.

Contact letterbox@leibniz.com.au with questions or objections before the stated date. We resolve reasonable data-protection objections before transferring affected school data, in accordance with the school’s agreement. Existing stricter residency commitments continue to apply.

AI marking is on unless a school turns it off. A school administrator can turn it off or back on for the whole school in Settings, and each change is recorded.

While it is off, students still answer by typing, handwriting or uploading their work, and it is saved for their teachers. Nothing they submit is sent to an AI model, so the work gets no marks or AI feedback. Multiple-choice questions are still marked automatically, without AI. Features that depend on AI marking, such as bulk submission and competitions, are hidden. Work that was already marked keeps its marks.

The current version of AI marking is 2.5 (8 October 2026). Typed answers and the key ideas in an answer are checked by open-weight AI models; handwritten and uploaded work is read and marked by a Google model. All requests go through OpenRouter to endpoints run by DeepInfra or Google that keep no copy of the data.

We change the version whenever the AI behind marking, or the way marking works for you, changes, and add the change below. Before switching on any new AI feature that uses student information, we give schools advance notice.

1. Version 2.58 October 2026
   • The teacher exam builder and its AI exam assistant were switched off for every account. Teachers can no longer build or edit exams with AI.
2. Version 2.44 October 2026
   • Schools can turn AI marking off. While it is off, students still submit their work and teachers can see it, but nothing is sent to an AI model and the work is not marked.
3. Version 2.330 September 2026
   • AI summaries of individual students for teachers, and AI-personalised tasks, were switched off. AI is now used only to mark work and to write practice questions.
4. Version 2.27 September 2026
   • When a marking request is retried, the retry also goes only to providers that keep no copy of student work.
5. Version 2.12 September 2026
   • Every AI marking request must come from a signed-in user or a short-lived demo session.
6. Version 2.025 August 2026
   • New Google model for reading and marking handwritten and uploaded work, and for matching uploaded pages to question parts.
7. Version 1.26 August 2026
   • Each marking task now uses one approved model. Requests go only to providers that keep no copy of the data and do not use it for training.
8. Version 1.14 May 2026
   • New model for checking the key ideas in an answer.
9. Version 1.0January 2026
   • AI marking of typed and handwritten answers, with feedback. This is the version in use when this history starts.