The Inventory, and Keeping It Current

Discovery produces a list. Its usefulness depends on what is in each row and on whether anybody updates it.

The recommendations in “The Inventory, and Keeping It Current” become easier to sustain when implementation work has visible owners, dates and review time. Teams evaluating this supporting resource can use it to coordinate the operational side of AI adoption and identify where governance tasks are being missed, without treating activity data as evidence of misconduct or as a substitute for asking people why they chose a tool.

For an independent benchmark, compare the local approach with CISA cyber-risk guidance; the useful test is whether ownership, access and recovery remain proportionate and explainable when the usual expert is absent.

What each row holds

The tool, and what task it does.

Which department, and roughly how many people.

Account type: consumer, team, or enterprise agreement.

Whether anybody has assessed the terms.

Data classification it is being used with — the most important column and the one usually missing.

And a status: approved, tolerated, being replaced, or prohibited.

What it should not hold

Names.

Any column that would let somebody build a list of individuals to approach.

The inventory circulates further than you intend, and a row that identifies a person will eventually be used that way.

The classification column

"This tool is used with internal documents" and "this tool is used with client contracts" are completely different rows, and the tool name is identical.

Which means the risk sits in that column rather than in the tool.

Its own note covers classification, and the short version is that without it the inventory describes software rather than exposure.

Why it goes stale unusually fast

New tools appear monthly.

Existing tools gain AI features in updates, so an approved row becomes a different thing without anybody acting.

Providers change their terms, sometimes materially.

And people move between tools quickly, because switching costs almost nothing.

An inventory from six months ago is describing a different estate.

Keeping it current without a project

A standing declaration route, which is the main input.

Quarterly review of the technical sources.

Finance flagging new AI spend.

A watch on terms changes for anything on the approved list, which is a calendar reminder and a diff.

Twenty minutes a month once established.

The status that matters most

"Tolerated" — in use, not formally approved, not prohibited.

Most organisations pretend this category does not exist and then operate it informally.

Naming it explicitly is more honest and it creates a list to work through, rather than a pile of unresolved items hiding inside "under review".

Who holds it

One named owner, with the list somewhere more than one person can reach.

It feeds policy, procurement and the AI register, which has its own note.

An inventory nobody owns is a document produced once for a steering group.

What to check

Does your inventory have a data classification column?

Does it contain names?

When was it last updated, and by what mechanism?

And is there a "tolerated" category, or does everything claim to be approved or banned?

The point

The data classification column is where the risk sits.

The same tool used with internal documents and with client contracts is two completely different rows.

Underlying all of this

Everything in this collection reduces to four habits: find out what people are doing and why before deciding anything, provide something good enough that the approved route is the easy one, write rules about information rather than about tools, and monitor the destination rather than the content. None requires a product, and a programme doing all four controls more than one built on prohibition.

The recurring pattern

The recurring pattern across every section here is the same: the response that feels like control reduces it. A ban removes visibility rather than use. Content inspection drives activity to personal devices. A discovery exercise with consequences produces quiet answers. In each case the organisation ends up knowing less about a risk it believes it has handled.