AI Policy

Cibbbo uses artificial intelligence to identify ingredients from photos, suggest recipes, parse imported recipes, and power the Ask Cubo assistant in text and voice. This AI Policy explains how those features work, how your data is used, and the limits you should expect.

By using Cibbbo AI features, you agree to this policy together with the Terms of Service and Privacy Policy. If you do not agree, do not use those features.

1. Where Cibbbo Uses AI

Fridge and pantry scans: photos you capture or upload are analyzed by computer-vision models to estimate ingredient names, quantities, and related details.

Recipe suggestions: models help match recipes to your inventory and food preferences. Allergy filtering runs on our own servers, not in the AI model; your allergy list is never sent to the AI provider.

Ask Cubo: an in-app assistant that answers cooking questions by text or live voice and can propose app actions (like adding items to a list) that you confirm.

Recipe imports: AI helps parse recipe pages you import from the web into structured ingredients and steps.

Age-appropriate answers: Cubo does not help users under 21, or users whose age is not confirmed, with alcoholic drinks, and offers alcohol-free versions instead. Recipes built around alcohol are hidden from them in the app.

2. Third-Party AI Providers

AI inference currently runs on OpenAI models via our Supabase backend. Providers may change as we improve the service; any provider processes your content only to deliver Cibbbo features.

Nothing is sent to an AI provider until you agree on the AI & your data screen. You can withdraw that permission anytime in Settings → Privacy Choices; AI features then ask again before sending anything.

Under our API agreements, our AI providers do not use your content to train their own models.

When Cubo looks up live prices, aisles, or deals, it uses our AI provider’s web search. We ask the provider not to keep those requests, and for deals we send only your city or ZIP code.

Over time, some detection may move to Cibbbo’s own models, trained only on data from users who opted in (see Section 4).

3. Voice Conversations & Transcripts

When you talk to Ask Cubo, your microphone audio is streamed to our AI provider in real time so the assistant can hear and respond. Audio is only captured during an active voice session.

Conversation transcripts may be stored with your account for context and history. Dictation (speech-to-text while typing) is handled on-platform by Apple under its own terms.

Cubo memory: Cubo keeps short notes about lasting preferences you mention (diet, household, favorite cuisines, equipment, skill) to personalize answers. See or clear them in Settings → Privacy Choices.

Thumbs up and down on Cubo replies are stored with your account and used only to tailor your own future answers, not to train AI models.

4. Photos, Corrections & Model Training

Scan photos are stored in private storage and sent to AI providers only to generate detections for your session.

Help improve Cibbbo’s scanner is off unless you turn it on, on the AI & your data screen or in Settings → Privacy Choices. When it is off, your corrections are not used for training and scan photos are deleted within 24 hours.

When you opt in, your corrections and scan photos may be reviewed by our team as part of a human-in-the-loop pipeline that improves detection quality. Training data is limited to opted-in accounts, and allergy information is never used for training. If you opt out later, your past corrections stop being used and your saved scan photos are deleted.

5. Accuracy & Your Responsibility

AI detections, transcripts, and suggestions are estimates. They can be incomplete, outdated, or wrong — including ingredient names, quantities, nutrition data, and allergen information.

Never rely on AI output alone for allergy, dietary, or food-safety decisions. Always verify ingredients, labels, and preparation yourself.

Cibbbo does not provide medical, nutritional, or food-safety advice.

6. Acceptable Use of AI Features

Do not use AI features to generate or spread harmful, deceptive, or infringing content, to attempt to extract other users’ data, or to probe or abuse the underlying models.

We may rate-limit, filter, or restrict AI features to keep them safe and available for everyone.

7. Changes & Contact

We may update this AI Policy as features and providers change. Continued use after changes means you accept the updated policy.

Questions: contact@veril.studio.