Leda by da.care SA — AI disclosure & consent

Version 1.1 (draft) — 2 July 2026

This is the in-app disclosure shown before you first use any AI feature, and is also available any time from Settings ▸ Data Controls and from the chat screen. It exists because Apple App Review (Guidelines 5.1.1(i) and 5.1.2(i)) requires this information to be presented in the app itself — not only in the Privacy Policy.

What is sent to AI providers

When you use Leda's chat, voice, or camera features, the following is sent to a third-party AI provider to generate your response:

• Your messages and questions (text prompts)
• The conversation history from your current session (so the reply has context)
• The active mini-app's instructions (which mini-app you're using and its persona)
• Your language preference
• Camera images — only when you use the camera/vision feature
• Voice recordings — only when you use the voice feature (transcribed to text, then a spoken reply is generated)
• A pseudonymous identifier used for rate-limiting and abuse prevention — not your name, e-mail, or account

What is not sent: your name, e-mail address, or account/user ID. AI requests are authenticated at Leda's own backend, not by the AI providers.

Training & retention: Your inputs are never used to train AI models. Providers may keep them for a short period for safety and abuse monitoring, then delete them. Your chat history stays in your account until you delete it.

AI providers we work with

Depending on the model in use, your messages may be processed by one of the following providers. Each has its own privacy policy:

• Anthropic (Claude) — Privacy policy ↗
• OpenAI (ChatGPT; Whisper for voice) — Privacy policy ↗
• Groq Cloud (serving Meta Llama) — Privacy policy ↗