AI Model Documentation
Version 1.0 · effective 2 October 2026 · last reviewed 2 October 2026 · next review 2 April 2027
In short
The AI models ReviseBase uses, which features use them, what is sent to them, and what we have and have not confirmed about data handling.
Read the full policy below.
This page documents the AI models ReviseBase uses, which features use them, and what is known about how each provider handles data. It is maintained by us and reviewed on the date shown at the top.
How ReviseBase uses AI
ReviseBase does not run or train its own AI models. Every AI feature sends a request through our application platform's AI integration, which routes it to a third-party model. Those models belong to three families: OpenAI's GPT models, Google's Gemini models, and Anthropic's Claude models.
Which model serves a request depends on your plan and on the feature. Paid plans reach more capable models where the platform makes them available.
What is sent
Only what a feature needs. That can include:
- text extracted from a study material you uploaded
- the notes and summaries generated from it
- the question you type in the tutor, and the recent conversation
- the text you submit for an originality check
- the URL of a web page or YouTube video you add as material
Your name, email address and payment details are never sent to an AI provider.
Model families
OpenAI, GPT
Used for quiz generation, flashcard generation, the AI tutor, processed study materials and note explanations on the plans routed to it.
- Privacy: https://openai.com/policies/privacy-policy
- Terms: https://openai.com/policies/terms-of-use
Google, Gemini
Used wherever a feature needs live web search: web and YouTube materials, originality checks, material verification, and Deep Research lessons with research enabled.
- Privacy: https://policies.google.com/privacy
- Terms: https://ai.google.dev/gemini-api/terms
Anthropic, Claude
Used for quiz generation, flashcard generation, the AI tutor, processed study materials and note explanations on the plans routed to it.
- Privacy: https://www.anthropic.com/legal/privacy
- Terms: https://www.anthropic.com/legal/consumer-terms
Platform-managed selection
One routing option is automatic, where the platform chooses the model per request. It is used as a fallback if a selected model is unavailable, so a request still completes.
What we have and have not confirmed
- We have recorded an official source for each provider's privacy policy, terms and, where available, data processing addendum. The links are on our Subprocessors and Third-Party Services page.
- We have not independently confirmed each provider's retention periods, training use or security certifications for the specific route our requests take. Those are marked "Not confirmed" in our internal registry.
- Our own compliance obligations, such as a signed data processing agreement with our platform provider and a confirmed transfer mechanism, are also open items. We do not claim to have completed them.
Limits of AI output
AI-generated content can be wrong, incomplete or out of date. Quizzes, flashcards, summaries and tutor answers are study aids, not authoritative sources. Always check against your course materials. See our AI and Educational Disclaimer for more.