Safety principles
- Trinity AI should teach from the learner's current skill context rather than inventing an unrelated path.
- Wrong answers are treated as evidence of a misconception, not as a reason to shame the learner.
- AI-generated explanations may be wrong and should be reviewable by an adult, teacher, or tutor.
- Formal accreditation, grades, diagnoses, and school placement decisions should not rely on AI output alone.
Launch status
The public demo and alpha surfaces show the intended tutoring pattern. Production learner accounts, school SSO, and formal credentialing are not claimed as fully live unless a later page says so.
Reporting concerns
Report unsafe, inaccurate, or confusing AI feedback to hello@math.foundation with the page, question, answer, and expected correction.