AI and Privacy: What Happens to What You Type in the Chat — and How to Control It
In Brief
Every sentence you input into the AI assistant is sent to the provider's server. What happens next depends on three factors: the plan you are using, your privacy settings, and what you choose to paste.
ANEzostylia Editorial TeamAI Based on facts and researchSupported by sources · 3 sources
Based on facts and research. Content grounded in studies, books, and recognized sources (WHO, APA, PubMed, universities). Without symbolic interpretations.
Chatting with AI may feel as private as a notebook. It isn't — but you have more control than you think if you know a few settings and guidelines.
Every sentence typed into the AI assistant is sent to the provider's server. What happens next depends on three things: the plan you are using, the privacy settings, and what you paste in yourself. Here’s what you need to know — without panic and without naiveté.
In free and individual plans, providers usually can use conversations to improve models, but they allow you to disable this (OpenAI: Settings → Data Management → "Improve Model for Everyone"; Google Gemini: Activity in Gemini apps; Anthropic: privacy settings). In corporate plans (Team/Enterprise, API), data no is by default used for training — this is one of the main reasons to pay for a business plan.
What to never paste
PESEL numbers, ID cards, credit cards, passwords, and 2FA codes — even 'just for formatting.'
Medical and financial data of others without their consent.
Documents covered by confidentiality (NDA agreements, client data) — outside of a plan with a data processing agreement.
A simple technique: before pasting, replace names with roles ('client', 'employee A') and amounts with approximates. The task content remains intact, and the risk disappears.
Chat History and Memory
You can disable saving history or use temporary chats (which are deleted after 30 days). 'Memory' features (the model remembers your preferences between chats) are convenient, but review them from time to time and delete what you don’t want to keep.
What the law guarantees
GDPR gives you the right to access your data, have it deleted, and object to processing — AI providers operating in the EU must accommodate this. AI Act (2024) imposes transparency obligations on general-purpose model providers, including notifying about generated content and training data; regulations are being phased in until 2026–2027.
Three Settings for Today
Disable 'model improvement' on private accounts.
Enable 2FA on your AI account — chat history often contains more sensitive data than email.
Use a business plan for working with client data or anonymize.
Card of the Day delivered to your email
Receive your Tarot, Dream, and Energy insights every morning. No spam, easy one-click unsubscribe.
Android · Google Play
Ezostylia on Your Phone
Download the free app from Google Play — tarot, runes, horoscope, Life Map, and AI divination always at your fingertips.
Yes — deleted chats are typically wiped from the provider's systems within 30 days (unless the law requires longer retention). You can also request data deletion under GDPR.
We only use strictly necessary cookies and local storage to operate the service (login, language, game state). We do not display any ads. With your consent, we will also enable:
Analytics (PostHog, USA) — anonymous page views, no session recording, no chat content.
Advertising — campaign measurement (OpenAI pixel): sign-up, purchase, game entry. No profiling of esoteric content.
You can change your choices at any time: Terms → Cookie settings. Further details in the Privacy Policy.
