Version 2026-09-28 · Last updated: 28 September 2026
13 Tarot is operated by IntoSignal Studio. This page describes how the service handles information. It is a product disclosure, not a certification of GDPR, PIPEDA, or other compliance regimes.
You do not create an account. We do not run a user profile or reading archive for you.
The site and API run on Cloudflare (Workers, static assets, and Durable Objects used for anonymous quota/budget counters). Traffic may be processed in Cloudflare’s global network.
When you use Interpret or follow-up chat, your question, locale, spread, and card choices are sent to our server so it can call AI providers and return a reply. Providers may include Groq, OpenRouter (which may route to models such as DeepSeek or NVIDIA Nemotron), and/or DeepSeek directly, depending on availability and configuration. Those vendors process the request under their own terms. We do not use your reading text to build marketing profiles in this app.
Before a request is sent to an AI provider, the service may screen question and follow-up text with automated word-list rules (for example to block sexual content involving minors, or to show crisis resources for self-harm language). Blocked text is not forwarded to the model for a reading. Screening is imperfect and not a substitute for human moderation.
Interpret may set a short anonymous session cookie (for example 13tarot_sid) for abuse protection and daily answer limits. That cookie does not store your questions or readings. Cloudflare Turnstile may run a security check in the browser. Quota and spend counters may be stored server-side against the anonymous session, not as a named account.
If you accept the analytics prompt, we load Google Analytics (GA4) to measure anonymous usage such as locale, spread type, and whether Interpret was used. We do not send your question text, AI replies, card IDs, or chat content. Google may set its own cookies when analytics is accepted. “No thanks” keeps analytics off. Clearing site data resets this choice (13tarot_analytics_consent).
Listen reads an AI interpretation aloud in the browser with Piper for English and French. The interpretation text is not uploaded to a speech service. The first Listen for a language downloads a public voice file into the browser; later plays can use that cached file. Speech software may also load from this site or a CDN, which receives ordinary request data such as an IP address, not the reading.
We use Google Search Console for site indexing and search performance. It does not receive your reading content from the app.
If you open a donation / tip link, payment is handled by Stripe on Stripe’s pages. Card numbers are not collected by 13 Tarot. Stripe processes payment data under Stripe’s terms. Tips support the project; they are not sold as paid readings or tax-deductible charitable donations unless separately stated by a registered charity (we make no such claim).
Questions, card selections, AI replies, and chat stay in browser memory for the current session and are cleared on leave or refresh (unless you share text yourself). We may store language preference (13tarot_lang) and analytics consent locally. Those keys are not reading content.
If you allow the camera, frames are processed on your device for card matching. They are not uploaded as a photo library to IntoSignal.
The page may load fonts or libraries from third-party CDNs (for example Google Fonts or jsDelivr). Those providers receive standard web request data such as IP address.
AI provider API keys are stored only on the server (Cloudflare Worker secrets). They are never exposed in the browser.
The service is not directed at children. Do not use it to generate sexual content involving minors.
We may update this page as the product changes. The version date above is the current published text.