Go home

Worldwide

AI Disclosure

What is authored, what a model generates, and where the model is wrong — for every user worldwide.

Effective 15 November 2026

1. Two kinds of lesson

Most catalog lessons are authored. A lesson from a page photo, “make it about me,” and Maya’s spoken voice go through a language model or speech synthesis at a provider (Google, Anthropic, OpenAI).

2. Limits

Models invent, miss the level, and mix facts. The parent is responsible. Maya is not any country’s ministry of education and is not a credential.

3. Maya is an AI

Maya is a drawn character who speaks with the help of artificial intelligence. She is not a person, not a human teacher, and no person reads the conversation as it happens. On first open, Maya tells the child so, in the child's language.

4. Report an answer

Next to each explanation Maya gives in a lesson there is a “Report this answer” button. The child or parent picks a reason (wrong, scary or unsuitable, something else), and Maya stops that explanation. The report reaches us with the lesson and question id, without the child's name, and is reviewed. You can also report at hello@maya-tutor.com.

5. Data to the model

Only what that request sends. We do not train a private model of ours on children’s lessons. Third-party providers are bound by their contracts, may keep a request for a limited time to detect abuse, and do not train on it; some are in the US.