Socrat
← Socrat

Privacy Policy

Version 2.1 · 14 November 2026. Website edition of 18.11.2026.

Socrat: the Telegram bot @socratesaiai_bot and the website socrat.surf. Personal data controller: the owner of the Socrat service (legal name, registration code and address to be added).

Data questions: @som_mos_oms on Telegram.

Contents

Google user data

Connecting a Google account is optional. If you connect one, Socrat asks Google for access to the following and uses it only to do what you ask in the chat:

What we keep. The access tokens, encrypted with AES-256-GCM, and, for mail, the messages the assistant has processed and the drafts it prepared. Whatever the assistant reads from Google at your request becomes part of your conversation with it (the answer and the result of the action) and is kept and deleted like the rest of the conversation: see sections 12, 15 and 16.

Who receives it. To write a summary, an alert or a draft, the necessary text is sent to the language model named in section 14, only on hosts that keep nothing after answering (zero data retention). We do not sell Google user data, do not use it for advertising, and do not use it to develop, improve or train generalised AI or machine-learning models.

People. No one on our side reads your Google data, unless you ask us to look into a problem and agree to it, it is needed for security (for example, investigating abuse), or the law requires it.

Socrat's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

Disconnecting. /disconnect_google (or /apps → Google) removes the tokens from our server at once; you can also revoke access in your Google Account under Security → Third-party apps and services. /delete_account deletes everything the bot keeps about you.

12. Data we process

13. Purposes and legal bases

We do not sell data, show ads, use advertising or analytics trackers, or train our own models on your data.

14. Who receives data

To carry out a request, the necessary part of the data goes to:

Requests to the language models that write the answers (main and fallback) go only to zero-data-retention (ZDR) hosts: the host keeps no request after answering. Speech recognition (Microsoft, Mistral), Perplexity embeddings and Google images currently work the same way. Without zero retention, under the providers' own policies: Voyage AI ranking, Recraft and xAI images, OpenAI text-to-speech, Cloudflare PDF parsing, search. Providers operate in various countries, including the USA. By tapping «Accept» you consent to these transfers.

Text recognition in photos and scans (OCR) and image background removal run on our server, without sending anything out.

15. Retention

16. Your rights and deletion

You can find out what data about you we process, get a copy, have it corrected, restrict or object to processing, withdraw consent (this does not make earlier processing unlawful) and request deletion. Complaints: the Ukrainian Parliament Commissioner for Human Rights; in the EU, your national data protection authority.

17. Other people's data

By sending other people's data (contacts, emails, business chats, groups) you confirm you have the right to do so and, where required, have told them. We process business chat data on your instructions.

18. Security

The vault, cards, and the keys and tokens of connected services are encrypted (AES-256-GCM), and access to the vault and cards is logged; connections to providers use HTTPS. No protection is absolute: if a breach affects you, we will tell you and, where the law requires, the supervisory authority.

19. Automated decisions

The bot makes no decisions with legal effects concerning you: the AI only answers and carries out your instructions.

20. Contacts

@som_mos_oms on Telegram

See also the Terms of Use.