At Revision Genie, we believe in transparent and ethical use of AI in education. This policy explains how we use AI technology and our commitment to protecting our users, especially young learners.
1.1 Education First: AI is used to enhance learning and understanding, not to entertain or distract.
1.2 Human Oversight: AI features are monitored and reviewed by our team.
1.3 Age-Appropriate Design: AI interactions are designed with the ICO's Age Appropriate Design Code in mind.
1.4 Accuracy and Quality: We maintain high standards for AI-generated content and clearly tell users that AI can make mistakes.
2.1 Provider: Revision Genie uses large language models hosted on Microsoft Azure OpenAI Service (primarily GPT-4 family models), and Azure OpenAI embedding models for retrieval. We also use Azure Speech Services for text-to-speech in some language lessons.
2.2 No Training on Your Data: Microsoft contractually commits not to use Azure OpenAI customer data (including your conversations, uploaded documents and prompts) to train its foundation models, and we do not separately train models on your data either.
2.3 AI Identity: AI tutors clearly identify themselves as AI. They are not human and may produce incorrect or out-of-date information.
3.1 AI Tutoring: AI tutors ("Genies") respond to your questions in chat, generate step-by-step lessons, run quizzes, and mark exam-style answers.
3.2 Personalised Content: AI generates practice questions, lessons and flashcards tailored to your level and curriculum.
3.3 Feedback and Marking: AI provides constructive feedback on exam answers, including visual marking via image analysis where supported.
3.4 Recommendations: AI helps suggest the next topic or activity in your learning path.
3.5 Custom Genies (My Genies): If you upload your own documents to create a custom AI tutor, AI uses retrieval-augmented generation (RAG) to answer questions grounded in those documents.
4.1 What It Is: Revision Genie includes a memory feature that allows the AI to remember a small number of stable facts about you across sessions (for example, "target grade 8" or "GCSE Combined Science, exam June 2027"). This helps the AI personalise responses without you having to repeat yourself.
4.2 What We Save: We save notes only when you explicitly ask, or when you volunteer a clear and stable fact relevant to your learning. We do not save names of classmates or teachers, school names or other sensitive identifiers.
4.3 Your Control: You can view, edit and delete all stored notes at any time at Settings → AI Memory. You can also tell the AI to forget something in chat, or wipe everything in one click.
5.1 Content Filtering: AI responses are passed through Azure OpenAI's content filters and our own additional safeguarding checks.
5.2 Safeguarding: Conversations that suggest a user may be at risk are flagged for human review and may be retained and disclosed as set out in our Safeguarding Policy and Privacy Policy.
5.3 Bias and Quality: We continually review AI outputs and update prompts and configuration to reduce bias and improve quality.
5.4 Clear Boundaries: AI clearly identifies itself as AI, and is restricted by system prompts from giving medical, legal, or financial advice.
6.1 Minimal Data Use: AI receives only the data needed to give a useful answer (your message, recent context, relevant curriculum, your skill level, and any memory notes for the current subject).
6.2 No Training: As stated in section 2.2, neither we nor Microsoft Azure OpenAI use your data to train models.
6.3 Secure Processing: All AI interactions are encrypted in transit.
6.4 Retention: AI conversations are not retained by default. See our Privacy Policy and Data Retention Policy for full details.
7.1 Per-Message Feedback: You can rate any AI response with a thumb up or down. Negative feedback can include a report so we can investigate.
7.2 Memory Control: You can delete any stored memory note at any time in Settings → AI Memory.
7.3 Conversation Control: Conversations are not retained by default; you can delete any uploaded files at any time.
7.4 Transparency: AI-generated content is clearly produced by the AI tutor. Lessons and quizzes generated on-the-fly are labelled as such.
8.1 Learning, Not Cheating: Revision Genie is designed to support learning. AI tutors are configured to teach concepts, explain answers and ask follow-up questions, not to write whole pieces of coursework on demand.
8.2 Teachers: Class teachers and school admins can see student activity and progress where their school has authorised this.
8.3 Schools' Own Policies: Students are expected to follow their school's academic integrity policy when using Revision Genie alongside formal assessments.
9.1 We continually update prompts, safety filters and AI model versions as the underlying technology evolves.
9.2 We collect and respond to user feedback, including thumbs-down reports.
9.3 We regularly assess AI performance against educational outcomes.
If you have questions about how we use AI or want to provide feedback, please contact us at support@revisiongenie.com. For more information about data protection, see our Privacy Policy and Children's Code Compliance pages.