How URYVA uses AI
Effective date: 1 October 2026
Last updated: 1 October 2026
This page explains in short what the Privacy Policy (section 5) says in full.
AI guides; you author
AI works as a guide, not the author. You decide what to keep, change, and make your own. Sorush, URYVA's AI guide, asks questions and suggests directions; you choose what to publish, and you are responsible for it. URYVA labels AI-generated and AI-assisted content.
AI can be inaccurate. Treat its output as a starting point, not as fact, and check anything important before you rely on it or publish it. Your Mindscape, and anything Sorush says, is not medical, psychological, legal, financial or career advice.
What uses AI
In the app's words: "Some features (Generate, Compose, Roadmaps, Mindscape) send your prompts and related content to third-party AI providers to work." In more detail:
- shaping and writing Folios and their images, narration and transcription, and Folio Flows;
- building a Galaxy from text or voice;
- your Mindscape: insights about your interests and way of thinking, built from your bio and interests, your answers, and the Folios you publish, bookmark and read;
- translating bios, Mindscape insights and titles for readers who use another language.
Who receives what
| Provider | What it receives | What for |
|---|---|---|
| OpenAI (USA) | Your prompts, thoughts, drafts and conversations while you compose. Folio content. Mindscape inputs and insights. Text or voice you use to build a Galaxy. Text to be read aloud. | Writing Folios and their images, narration and transcription, Folio Flows, Galaxies and Mindscape |
| Anthropic (USA) | Profile bios, Mindscape insight texts, Folio and Galaxy titles and labels to translate, and some Folio story text. | Translation, and writing parts of Folios |
| xAI (USA) | The text of a Folio to be read aloud, only when the optional "Grok" narration voice is used. The current web app does not use it. | Voice narration |
To research a topic, URYVA also sends search queries that the AI writes about that topic to Brave Search (or DuckDuckGo), Wikipedia and Wikimedia Commons, and the World Bank. These queries do not include your name or your account details. The name, e-mail address and photo we receive from your sign-in provider are not sent to AI providers.
By default, under their API terms, these providers do not use the data to train their models: OpenAI does not unless the customer opts in, and keeps API data for up to 30 days for abuse monitoring; Anthropic does not by default; xAI does not without explicit permission, and deletes it after 30 days.
Your AI setting
The first time you use an AI feature, URYVA asks for your permission. You can turn AI processing on or off at any time in Settings → Privacy Policy → AI & Data. When it is off, Generate, Compose, Folio Flows, and the Mindscape refresh and questions do not run.
What the setting does not cover today
Some AI processing still happens when the setting is off:
- your Mindscape may update automatically after you publish or bookmark a Folio or edit your profile;
- building a Galaxy from text or voice uses AI;
- your bio and Mindscape insights may be machine-translated when someone who reads another language views them.
More
- Subprocessors: every service provider, with the data it receives and where.
- Privacy Policy, section 5: the full text, which prevails.