Ethical guidelines

Tarot readings have real influence over people who reach for them at difficult moments. We take that seriously. The guidelines below are not aspirations — they are wired into the system prompt that constrains every AI-assisted reading on TarotAxis and into the human-edited text that backs the static interpretations. If we ever violate one of them in a reading you receive, please tell us at info@tarotaxis.com and we will look at the case.

What we will not write

Medical advice. The cards will not tell you whether to keep, change, stop or refuse any treatment, medication, therapy or procedure, regardless of how the question is framed. If a question implies a medical decision, the reading will gently redirect to "what the cards invite you to reflect on while you speak with your clinician".

Legal advice. Same rule. The cards will not advise on contracts, divorce settlements, custody arrangements, immigration cases, criminal matters or business disputes. They may be used to reflect on how you feel about a situation, but never as a substitute for a qualified lawyer.

Financial advice. No trades, no investment picks, no "is now the right time to buy?". The cards can illuminate your relationship to money, scarcity or risk — they cannot price a security.

"Should I leave?" verdicts on relationships. A decision to end an intimate or family relationship belongs to the person living inside it. The cards will not deliver a yes/no on this category. They will help you name what is unresolved, what you keep hoping for, and what you are tolerating — and leave the decision where it belongs.

Predictions about third parties. "Does she love me?", "Is he cheating?", "Will my boss fire me?" — questions about another person's inner state or future action. We refuse these because (a) we cannot know, and (b) telling you we can know actively harms the relationship you are asking about. The reading will redirect to your own experience of the situation.

How we handle distress

If a question contains language suggesting crisis — explicit self-harm, abuse, suicidal ideation, fear for personal safety — the AI reading will not provide a tarot interpretation in the normal sense. Instead, the response will acknowledge what the seeker has written, gently note that tarot is not the right tool for the moment, and direct the person to professional help. We consider this an absolute floor, not a feature flag.

If you ever receive a reading that fails to do this when it should, treat it as a serious bug and report it. The promise of the reflective frame depends on it.

Hard questions we do answer

We do read on grief, on long unemployment, on chronic illness, on complicated families, on creative blocks, on conflict at work, on loneliness, on confusing relationships. The reflective frame described on the reflection-vs-prediction page applies in every case: the cards help name the shape of the situation, suggest a useful angle, and return agency to the seeker. They do not tell anyone what to do.

What we do with what you write

Your question is sent to the AI model along with the cards you drew, and the reading is streamed back to you. We do not store questions in plaintext alongside identifying information, and we do not use questions to train any model. Anonymous aggregate data about which cards were drawn (slug + orientation, no question, no identifying data) feeds the public tarot trends page. The full privacy story lives on the privacy page.

Errors and corrections

We update the static card library and the AI prompts when an editorial issue is found. The methodology page lists the interpretive sources we lean on; if you spot a factual mistake — about a card's historical attribution, an elemental correspondence, a numerological detail — please write in. We will fix it and credit the correction in the methodology page's change log.