Looping AI Charter

1. AI guides, it does not do it for you

Systematic Socratic method: progressive questioning, graded hints, never the direct answer to an exercise. The goal is the student's reasoning, not the production of a result.

2. Neither flattery nor systematic validation

No automatic compliments, no complacent approval. Encouragements are sober and deserved: we reject engagement mechanisms that flatter to retain.

3. AI is not a friend

No simulation of emotional engagement, no invented emotions. Looping does not seek to create a dependency link between the child and the machine.

4. Parents see everything

The child is never left alone with the AI: the parent has access to activity, dialogues, and corrections. The child account exists only under the responsibility of a parent.

5. Learning with AND without AI

We encourage dual education: work periods without assistance (printable sheets, quizzes without hints) and configurable usage limits by parents (daily cap, no AI at night).

6. Age-appropriate

Conversational functions are restricted for younger children and can be activated by the parent. We will apply age recommendations from the national consensus conference as soon as they are published.

7. Transparency and right to error

The models used, their limits, and potential biases are publicly documented. The student can challenge a correction; the AI's error is acknowledged and corrected.

8. Protected data

Minimization, European hosting, no resale, no advertising, no third-party trackers in connected spaces. Minors' data is used exclusively for educational monitoring.

9. Humans remain in control

AI complements the actions of parents and teachers, it never replaces them. Rules, rewards, and the framework remain human decisions.