Which of Your Own Data Categories Can Go Into AI, and Who Decides?

Data Classification Guide

Free PDF Updated The Jonesi Consultancy
Data Classification Mapping Worksheet For IT & Compliance Landscape
Data Classification Guide cover image
Download the Free PDF

Free. No email list, no follow-up sequence.

About this resource

What this guide covers

What Not to Put Into AI gives your staff the rule in one page. This is the exercise that produces the rule: a landscape worksheet for mapping your organization’s own data categories against permission levels and naming who decides for each one.

Generic data rules fail because every organization holds different things and calls them different names. The worksheet walks a group through listing their actual categories, assigning each a permission level, naming the approval authority where one is required, and recording the reasoning so the classification survives the person who made it. The output is a mapping specific enough to be enforceable, which a borrowed rule never is.

A generic data rule fails because your organization does not call its information what the template calls it. The mapping has to use your own category names.

How to use it

Purpose, and how this will help

Complete it before you publish a data rule, not after. Do it with someone who actually handles the information rather than only with policy owners, because the categories that cause incidents are the ones leadership does not know exist. The output feeds directly into the one-page rule your staff will carry, which is where it becomes useful.

Inside the download

What you'll take away

1

List your real categories

Your own names for your own information, not a template’s.

2

Assign permission levels

Never, approved tools only, or generally fine, category by category.

3

Name the approver

Who decides when a category needs a decision, by role.

4

Record the reasoning

Why each classification was made, so it can be revisited rather than re-argued.

Who this is for

  • IT, security, and compliance leaders producing a data rule
  • Operations leaders who know what information the organization actually holds
  • Governance owners completing the data question
  • Regulated organizations with existing classification schemes

Common questions

Frequently asked

How do you classify data for AI use?

List your organization’s actual information categories in the language your staff uses, assign each one a permission level, name who approves the cases requiring a decision, and record why each classification was made. Borrowed classifications fail because they describe somebody else’s information.

Who should decide what data can go into AI tools?

The classification belongs to whoever owns the data domain, with security and compliance confirming rather than deciding. Approval authority for edge cases should be named by role and reachable quickly, since an approver who takes a week is functionally a prohibition.

What if a data category does not fit any permission level?

That is the most valuable output of the exercise. Edge cases are where incidents originate, and surfacing them on a worksheet means deciding deliberately rather than leaving each employee to decide alone under time pressure.

Download the Free PDF

Free. No email list, no follow-up sequence.

Keep Reading

Related Resources

More from the resource library, chosen because they pair well with this one.