Security, Privacy & AI Oversight

This page covers what a review actually asks about: where processing happens, what the AI gets to see, when data is deleted - and the points where a human always decides.

• Servers in Germany, moderation AI inside the EU
• AI-written replies never go out unchecked
• Comment content is deleted after 30 days
• Uncertain cases end up with a human

The EU AI Act asks for human oversight of automated decisions. In Doorman that is not a setting somebody can forget, it is how the product is built: wherever the AI is uncertain or writes freely, a human decides.

Two rules always hold: if the AI fails or answers unclearly, the comment counts as needing review - never as silently approved. And every hide can be undone.

The application and the database run on servers in Germany. Moderation - the one step that touches every single comment - runs at an AI provider inside the EU.

What that means for the rest

• Reply drafts and the processing of your knowledge base run at a US provider, covered by EU Standard Contractual Clauses.
• Sign-in and payment run through established providers - passwords and payment details are held there, to their standards, rather than with us.
• Uploaded knowledge documents are stored under EU jurisdiction.
• The full list of providers and a data processing agreement are available on request.

The comment text goes out for checking - and nothing else that leads back to a person. Who wrote the comment is not needed for the decision, so it is not sent.

Sent to the AI

• The comment text
• Your rules, in your own words
• Context of the post that was commented on

Deliberately left out

• The commenter’s name or profile id
• Your Instagram or Facebook credentials
• Email address, subscription and payment data
• Other customers’ data - a request only ever holds yours

Deletion is a scheduled operation, not occasional housekeeping. What remains is a set of numbers with no personal reference - that is what your analytics are built from.

Delete your account and every piece of personal data goes with it, including uploaded files and stored credentials. A single channel can be disconnected and purged without giving up the account.

Access to your channels is limited to what moderation needs - and every route to it is secured.

Safeguards

• Access to your channels is stored encrypted (AES-256), with the key kept out of the database.
• Sign-in with a second factor; deleting the account asks for it again.
• Every connection runs over HTTPS.
• Incoming messages from Meta and the payment provider are signature-checked.

Access to your channels

• Doorman reads comments, hides them, makes them visible again and replies to them.
• Access stops at the comments: posts, direct messages and ad accounts stay out of reach.
• Only on the channels you connected yourself - disconnect whenever you want.

• The EU AI Act and automated comment moderation - deadlines, Article 50 and human oversight in detail.
• GDPR in automated comment moderation - legal bases, data minimisation and data subject rights.
• EU inference and the provider split - where each step of the processing runs.
• Security architecture and Meta permissions - scopes, webhook signatures and token encryption.

If your review needs a detail that is not on this page, write to us. You will get a concrete answer, not a form of words.

• Ask a security question
• Read the privacy policy

Nordify UG (haftungsbeschränkt), Hamburg. Last reviewed: 7 October 2026.