Security & Trust

Read-only first. Nothing changes without your approval.

Guvern looks at who can reach your content, so it has to be trustworthy itself. This page sets out what we ask for, what we read, where the data goes and how long we keep it.

How access works

  • Guvern is a multi-tenant app registered in Microsoft Entra ID, hosted in Distributed Logic's own tenant.
  • One of your Microsoft 365 admins approves it. That creates a service principal for Guvern in your tenant, which you can see and manage like any other enterprise app.
  • Your people sign in with their existing Microsoft 365 work accounts. There are no new passwords.
  • You can withdraw consent at any time: in the Entra admin center, open Enterprise applications, find Guvern and delete it. Guvern's access ends at once.
  • If your tenant only lets admins approve apps, users see Microsoft's "Need admin approval" page. That is expected; an admin approves once for everyone.

What each package asks for

Each package asks only for the access it needs. You see the full list on Microsoft's consent screen before you approve.

PackageAccessPermissions
Free readiness scan and Copilot governance checkRead-onlyShown on Microsoft's consent screen before anything is granted
Copilot Readiness AssessmentRead-onlyShown on Microsoft's consent screen before anything is granted
Copilot Usage Analysis and MonitoringRead-only, usage and license reportsShown on Microsoft's consent screen before anything is granted
Conversation Analysis add-onExtra read access to Copilot interaction recordsShown on Microsoft's consent screen before anything is granted
Custom RemediationWrite access, granted only for the job you approved, then removedShown on Microsoft's consent screen before anything is granted

What we read, and what we never read

We read

  • Sites, libraries, folders and file names
  • Permissions, sharing links and group membership
  • Guest accounts and their settings
  • Sensitivity labels and activity dates
  • Copilot settings, licenses and usage totals

We never

  • Open or copy the contents of your files during a scan
  • Change a setting or permission without an approved remediation job
  • Read mail or chats for the scan or assessment
  • Sell or share your data with anyone

Conversation Analysis is the one exception to "no conversations": it reads Copilot prompts and responses, only when you buy it, and reports results in aggregate.

Where your data lives and how long we keep it

  • Scan data is processed in Microsoft Azure, in the Azure region we use, which we will tell you on request.
  • We build your reports, deliver them, then purge the raw scan data. Every purge is written to a deletion log you can ask to see.
  • Reports stay available in your Guvern portal for as long as your service runs, or until you ask us to delete them.
  • Planned: keep usage history and reports in your own Azure subscription, or run Guvern entirely inside your tenant.

Safe remediation

When you buy remediation, every fix follows the same four steps. Nothing in your tenant changes until your approver says yes.

1 Simulate2 You approve3 We run it4 Before/after report
  • Simulate: we run the fix in simulation mode and list exactly what would change.
  • Approve: your named approver reviews that list and approves or rejects it.
  • Run: we run the approved fix on exactly the items in the simulation, nothing more.
  • Report: you get a before and after record of every change.

How we use AI

  • We use Anthropic's Claude to turn raw findings into a prioritized, plain-language report.
  • Your data is not used to train AI models.
  • A Distributed Logic consultant can review any AI-written report before you rely on it.

Compliance roadmap

  • Microsoft publisher verification: planned.
  • Microsoft 365 App Compliance certification: planned.
  • SOC 2: on our roadmap after launch.
  • A data processing agreement is available for every paid customer. See Terms & DPA.

Report a security issue

If you think you've found a security problem in Guvern, email info@distributedlogic.com with "Security" in the subject. We reply promptly and keep you updated until it's fixed.