AI Safety Disclaimer
Last updated: July 9, 2026
Mendoria uses large language models for its mentor conversations, journal reflections, and plan generation. These systems are genuinely useful for reflection — and genuinely limited. Here is what we do about that, and what remains yours to hold.
What the AI is
A language model generates plausible text from patterns. It does not know you, does not remember beyond what the product provides it, has no professional training, and holds no responsibility. Its answers can be inaccurate, generic, subtly biased, or confidently wrong.
Boundaries we build in
- The mentor is instructed never to diagnose, prescribe, or present itself as a therapist or doctor — and to encourage professional support when things sound heavy.
- Crisis handling: messages and journal entries pass a safety check before any AI model sees them. Content suggesting acute distress receives a built-in supportive response with real helpline information instead of a generated reply — and when the check is uncertain, we err toward showing support resources.
- The mentor is instructed to avoid manipulative or dependency-forming language, absolute claims, and fear-based motivation.
- Daily usage limits keep the relationship a tool, not a tether.
What no safeguard can promise
No automated check catches everything. A generated response may still miss your context, land poorly, or fail to recognize distress. If a conversation ever leaves you feeling worse, pause it — and if you are struggling, bring it to a human: a professional, a helpline (findahelpline.com), or someone you trust.
Your part
Treat AI reflections as prompts for your own judgment, never replacements for it. You are the authority on your life; Mendoria is a mirror, and mirrors have distortions.