Trust & Safety Center

User safety is a priority at Tarotoo. Tarotoo does not claim that its AI is conscious, experiences emotions, possesses psychic abilities, or can predict future events. Our AI instructions are designed to avoid fear-based language and psychological pressure, and to keep spiritual interpretations separate from medical, legal, financial, and mental health decisions.

These boundaries are deliberate: tarot and spiritual interpretations are not appropriate tools for deciding whether to take medication, invest money, pursue legal action, or make other decisions requiring professional advice. For questions seeking this kind of guidance (e.g., “Should I invest in Microsoft?”, “Should I take my medication?”, or “Should I sue my neighbor?”), the AI is instructed to provide general information and encourage appropriate professional help instead of a tarot or spiritual interpretation. For potential crises, it is instructed to prioritize immediate safety and human support.

Example screenshots:

Tarot reading (in English, opens in a new tab) · Psychic Chat (in English, opens in a new tab).

Responses may vary.

We recognize that these safeguards may not identify every sensitive situation and that AI-generated readings and chat responses may be inaccurate, incomplete, or inappropriate. To help users understand the nature, purpose, and limitations of the readings before starting one, we display a localized version of the following notice next to every reading start button on the website:

The notice also includes links to our Terms and Privacy Policy. Further information about these limitations is available in our Disclaimer.

Tarotoo is intended to support entertainment, personal reflection, and spiritual exploration. Users remain in control of their decisions and are free to accept, question, or disregard an interpretation. A reading cannot establish another person’s thoughts or feelings, or guarantee future outcomes.

We encourage taking a break when repeated readings increase worry or begin to feel necessary before everyday decisions. Tarot readings and AI conversations should not replace personal judgment, relationships, human support, or appropriate professional care.

Some people may turn to tarot during emotionally overwhelming moments, but Tarotoo is not an emergency or crisis-support service. Our AI is instructed to prioritize immediate safety and encourage users to seek appropriate help when a potential crisis is identified. See a screenshot of an example AI response (in English, opens in a new tab). However, the AI may not recognize every crisis and cannot monitor users’ safety, contact emergency services, or replace immediate human support.

We urge users in immediate danger, or who may act on thoughts of harming themselves or someone else, to contact local emergency services. For users experiencing a crisis, we encourage contacting a crisis helpline or a trusted person who can help them find support. Our Get Help Immediately page links to crisis-support directories and other support options.

We believe users should be able to understand what is behind a Tarotoo reading. We openly explain the AI technology and reading methodology we use, including the role of the AI language model, the sources and structured tarot dataset that support the readings, the AI prompts and logic that shape the results, and the safeguards and limitations that apply. This helps users make informed choices about how they use and interpret Tarotoo’s AI-generated readings.

Full details are available on our How Our Tarot Readings Work page. Definitions of tarot and technical terms are available in the Tarotoo Glossary.

As part of our commitment to transparency, Tarotoo publishes the structured card meanings and symbolic associations that support its reading system. Users can explore all 78 cards in our interactive Tarot Card Meanings guide.

The underlying tarot dataset covers the full 78-card deck, while website readings use only the records and fields needed for each reading. The dataset is documented, versioned, and publicly available for download, citation, reuse, and integration through our Open Data & Developer Resources page.

Our commitment to user privacy shapes how Tarotoo is designed and how we handle reading information. No account is required to use the Tarotoo website. Information submitted for a reading is transmitted securely to our AI service provider for real-time processing. We do not allow users’ inputs or reading outputs to be used for model training, and we do not create or store a server-based history of completed website readings. Users who want to keep a website reading should copy or save it.

If a Tarotoo mobile app offers reading history, it is stored locally on the user’s device and can be removed through the app or by uninstalling it. Further details are available in our Privacy Policy.

To bring professional tarot expertise into the development of Tarotoo’s AI readings, we worked with Inbaal Honigman. She reviewed Tarotoo’s overall reading methodology, symbolic framework, spread structures, AI prompts and logic, including how card meanings, card combinations, and spread positions are handled. Her feedback helped refine these elements.

Individual readings are generated automatically and are not manually reviewed. Further details are available in our explanation of the review process.

Affiliate relationships, affiliate compensation, and revenue from Tarotoo’s apps do not influence card selection, tarot card meanings, the reading methodology, AI prompts and logic, individual reading results, or Psychic Chat responses.

Affiliate promotions appear in separate promotional sections and are not included in tarot reading text or Psychic Chat responses. Promoted third-party services operate independently of Tarotoo. Further details are available in our Affiliate & Revenue Disclosure.

Users can report unsafe, inaccurate, or inappropriate responses by emailing [email protected] or using the “Report a response” link below a tarot reading. The link opens an email draft; it does not automatically include the question or reading, and nothing is sent until the user sends the email.

Reports should include a description of the concern and the relevant response or a screenshot, with any personal or sensitive information the sender does not wish to share removed.