NAV and Business Central blogs from Innovia

AI Governance Starts with Your Data: How to Protect Information While Expanding AI Adoption

Written by David Brown | 20 Nov 2026

AI is already part of how people work. Employees are using it to draft content, summarize information, analyze data, prepare presentations, and answer questions faster.

The challenge is that not all of that AI use is happening within company-approved tools or accounts.

That creates a growing governance question for businesses: Where does your company data actually go when employees use AI?

The biggest risk often isn’t AI itself. It’s a well-meaning employee putting company information into a tool or personal account that the organization hasn’t reviewed, approved, or configured.

The goal shouldn’t be to prevent people from using AI. It should be to give them a safe way to use it.

Your AI Account Matters

One of the easiest details to overlook is the account employees use to access AI.

The same application can provide very different protections depending on whether someone signs in with a personal account or an organization-managed business account.

With personal AI accounts, companies may have little visibility into what employees share, how the provider handles that information, or what happens to it afterward. Business and enterprise plans generally provide stronger centralized controls, although the exact protections depend on the provider and subscription tier.

A strong first step is simple: provide approved business accounts and manage their settings centrally.

Before approving an AI platform, organizations should also understand:

    • Whether company data can be used for model training
    • How long prompts and responses are retained
    • Whether vendor employees can review conversations
    • Where information is processed and stored
    • Who inside the organization can access logs and change settings

These questions should be answered for the specific subscription and configuration your company plans to use.

AI Can Expose Existing Permission Problems

Moving to enterprise AI solves some problems, but it can also reveal another one: oversharing.

Microsoft Copilot works within a user's existing permissions. It doesn't automatically give someone access to information they couldn't previously open.

But AI makes finding information significantly easier.

Imagine that a sensitive spreadsheet was accidentally shared with a large group several years ago. Employees might technically have access without knowing the file exists. Copilot can make that information much easier to discover through a simple question.

Before expanding Copilot access, review:

    • SharePoint sites and folders available to large groups
    • Old department and project sites
    • Files shared externally
    • "Anyone with the link" permissions
    • Groups that provide broader access than necessary

This isn't just an AI cleanup project. It's an opportunity to address access issues that may have existed long before AI arrived.

Don't Forget About Business Central

For companies using Microsoft Dynamics 365 Business Central, permissions deserve particularly close attention.

Copilot works on behalf of the individual user and follows that user's Business Central permissions. If users have broad permission sets, AI may be able to work with the same broad range of information.

AI agents require the same kind of consideration. Each agent should receive only the permissions necessary to perform its intended job.

Businesses should review Business Central's Copilot and agent capabilities, user permission sets, external AI connections, and who currently holds highly privileged access such as SUPER or D365 FULL ACCESS. 

Put Simple Rules Around AI

Technology controls aren't enough on their own. Employees also need clear guidance about which information belongs in which tools.

A straightforward classification might look like this:

    • Public: Published information that can be used broadly
    • Internal: Routine company information that should stay within approved tools
    • Confidential: Customer lists, contracts, pricing, and similar information that should only be used within approved enterprise tools
    • Restricted: Highly sensitive financial, personal, payroll, health, or payment information that requires specific approval before being used with AI

Keep the rules easy to find and easy to understand. Employees should know which AI tools are approved and which account they're expected to use before they have to make that decision themselves.

Three Places to Start

You don't need to build an extensive AI governance program overnight. Start with three practical steps:

1. Find out what employees already use.
Identify AI tools already being used throughout the organization and establish an approved list.

2. Review the controls you already have.
Look at Microsoft 365 Copilot settings, SharePoint permissions, Microsoft Purview capabilities, Business Central permissions, agents, and other controls already available within your Microsoft environment.

3. Move work to managed accounts.
If employees are already finding value in AI, give them approved business accounts where your organization can centrally manage access and settings.

AI adoption is going to continue. The organizations best positioned to benefit from it will be the ones that understand where their information goes, provide employees with appropriately configured tools, and establish rules people can realistically follow.