diff --git a/security.md b/security.md
new file mode 100644
index 000000000..ff35cd9b0
--- /dev/null
+++ b/security.md
@@ -0,0 +1,45 @@
+# Security policy
+
+We believe in transparency to mitigate security risks. All known vulnerabilities are available on 
+our [security page](docs/modules/ROOT/pages/security.adoc).
+
+We disclose such security issues only once a released version addressing the issue is available.
+
+We use automated tools to review our docker images and dependencies.
+
+# Reporting security vulnerabilities
+
+To ensure safety of our users, security process needs to happen privately.
+
+Here are the steps:
+
+ - 1. Reporter email the issues privately to `openpaas-james[AT]linagora.com`.
+ - 2. We will then evaluate the validity of your report, and write back to you within two weeks. This response time 
+ accounts for vacation and will generally be quicker.
+ - 3. We will propose a fix that we will review with you. This can take up to two weeks. 
+ - 4. We will propose a draft for the announcement that we will review with you. 
+ - 5. We will propose you a schedule for the release and the announcements.
+ - 6. One week after the release we will disclose the vulnerability.
+ 
+You will be credited in the vulnerability report for your findings.
+
+# Threat model
+
+The following threats are generic points of attention for email softwares:
+
+ - Virusses: Emails are one of the vector for spreading virusses. We recommend administrators to set up virus scans
+as part of their email infrastructure. We recommend user to be cautious opening attachments of unknown senders or
+suspicious emails. We recommend users to have an anti-virus installed. For instance [TeamMail backend](https://github.com/linagora/tmail-backend/) is integrated with 
+[ClamAV](https://www.clamav.net/) solution.
+ - Fishing: Attackers can try trick users on their identity and try to make them believe they are a legitimate sender and try to use
+this to either make user conduct actions or leak information. We recommend administrator to run an anti-Spam software, to verify SPF and DKIM
+records. For instance [TeamMail backend](https://github.com/linagora/tmail-backend/) is integrated with [RSpamD](https://rspamd.com/) solution.
+ - `Authentication`. We recommend administrator to set up strong authentication with an OIDC provider. This avoids TeamMail frontend to store directly user credentials,
+enable configuring handy features like two factor authentication and the like. For instance [TeamMail backend](https://github.com/linagora/tmail-backend/) is integrated
+with [LemonLDAP](https://lemonldap-ng.org/) identity provider using the [Apisix](https://apisix.apache.org/) API gateway. We also recommend users to logout once they finished using TeamMail.
+ - HTML rendering. By design, email embeds HTML generated by non trusted sources. TeamMail do sanitize HTML prior displays. The use of the canvas also limits the 
+impact of HTML rendering related vulnerabilities. Loading of remote resources like images can also be used for tracking purposes. As off today, TeamMail do not allow 
+blocking such tracking attempts.
+
+TeamMail today relies on [Firebase Cloud Messaging](https://firebase.google.com/docs/cloud-messaging) for its push architecture. 
+Only StateChanges are transiting though a third party, which to not include presonnal data.