---
title: "Security & trust: AI keeps track, your rules decide"
description: "Storage in the EU, review before anything reaches everyone, access per team, app changes that wait for a person, an emergency stop and an audit log."
url: https://wemember.ai/security
updated: 2026-10-14
---

Security & trust

# AI keeps track. Your rules decide.

Your people already put company information into AI tools. Wemember gives that a route you control: per team, per app, reviewed before it reaches everyone.

[Connect your AI](https://wemember.ai/apps#connect)

Storage in the EU · No per-person tracking

Trust

## Six things that are always true.

Whatever AI tool your team uses, these hold.

**Storage in the EU**

Your company memory is stored in the EU.

**Reviewed before it reaches everyone**

Company-wide knowledge, procedures, sharing and app actions are proposed first. Personal and team notes are applied and you are notified.

**Never more than the person**

An agent works with the rights of the person it works for.

Fixed

**App changes wait for a person**

Every change in an app waits for approval. This is fixed and is not a setting.

**Emergency stop**

One switch sets every agent back to propose-only.

**Audit log**

Every event is logged and exportable as JSONL.

Rules

## The server enforces it, not the AI tool.

An agent can't widen its own rules; only an admin can, and that is logged. Per action and per scope you choose: propose, or decide. Undo stays possible. A second model check is always on.

People

## Roles that match how you work.

Members use and propose within their teams. Reviewers publish knowledge and approve actions. Admins change the rules, the apps and the people.

Settings

## What is fixed, and what is yours to set.

A few things are not settings. Everything else is a switch an admin can see and change.

### Fixed. Not a setting.

**Every change in an app waits for a person**

Every action that changes something in your apps waits for a person's yes. That rule is built into the server; an agent can't approve it itself and there's no setting to turn it off.

**An AI tool never sees more than the person who connected it**

Access follows the person and their teams.

**Storage in the EU**

Where your company memory lives.

### Yours to set.

**Propose, or decide**

Per action and per scope. The safe default is propose.

**Who reviews**

Reviewers publish. Admins change the rules, the apps and the people.

**Emergency stop**

Sets every agent back to propose-only. Reading keeps working.

Switch positions are illustrative.

For your privacy officer

## Questions we expect, and answer in the call.

Short answers here. The full answers, with the processing agreement, in a call with your privacy officer.

[Bring your questions to the call →](mailto:hello@wemember.ai)

**Where is data stored?**

Storage in the EU.

**Processing agreement and GDPR questions?**

Covered in the call.

**What happens with Gmail or Drive data?**

Only what the rules for that app allow.

**Can an AI tool read more than its user?**

No. An AI tool never sees more than the person who connected it.

**Can an agent change something in an app on its own?**

No. Every change in an app waits for a person. This is fixed, not a setting.

**How do we audit this?**

Every proposal, decision and change is logged, with who and from which tool.

**What if we need to stop everything?**

The emergency stop sets agents back to propose-only. Reading keeps working.

## One memory for every AI your team uses.

Add one address to your AI tool. Your rules decide what it may read and propose.

[Connect your AI](https://wemember.ai/apps#connect)

Storage in the EU · Reviewed before it reaches everyone · Access per team