Platform Banner

Shadow AI: Identifying Uncontrolled AI Use Tutorial

Shadow AI is the use of artificial intelligence tools without approval, a framework, or visibility. The video explains how tools slip into daily work, with examples like pasting customer emails for summaries or generating visuals, and questions data access, confidentiality, reliability, and copyright. It suggests asking: Is it authorized? What data is used? Is human validation necessary, and how is it governed by the AI Act?

02:21
14 views
Share

Available in AI - Artificial Intelligence

00:00:07

we tend to think first of official projects,

00:00:10

a tool approved by the IT department,

00:00:11

a solution reviewed by the legal team,

00:00:13

or a project led by a business unit.

00:00:16

But AI is also spreading in another way,

00:00:18

more discreetly,

00:00:19

more quickly,

00:00:20

and often with less control.

00:00:22

Tools are making their way into everyday work

00:00:24

without ever going through an approval process.

00:00:27

This is what we call shadow AI,

00:00:29

the use of artificial intelligence tools without approval,

00:00:32

without a clear framework,

00:00:33

and without real visibility for the organization.

00:00:37

Most of the time,

00:00:37

there's no bad intention behind it.

00:00:39

The goal is simply to save time,

00:00:41

summarizing a note,

00:00:43

generating a visual,

00:00:44

preparing for a meeting,

00:00:45

or developing an idea.

00:00:47

The problem is that a very practical use can raise very concrete questions,

00:00:51

especially when it comes to data.

00:00:54

Take pasting a customer email into a publicly available AI tool,

00:00:57

for example.

00:00:58

Where does that data go

00:00:59

and who can reuse it?

00:01:01

Or asking it to summarize an internal document?

00:01:04

Am I sure it contains nothing confidential?

00:01:06

The same applies to a visual or a text intended for external use.

00:01:10

Is the content reliable,

00:01:12

free of copyright restrictions,

00:01:13

and approved for that purpose?

00:01:15

Generated content is not automatically publishable content.

00:01:19

This is where the AI Act helps clarify things.

00:01:22

The regulation does not declare AI dangerous,

00:01:25

nor does it prohibit its use.

00:01:27

It encourages us to look at how it is used,

00:01:29

the context and the consequences,

00:01:31

exactly what is missing with shadow AI.

00:01:34

The real risk of shadow AI lies in its invisibility.

00:01:38

When these uses multiply without a framework,

00:01:40

no one really knows which tools are being used,

00:01:42

what data is circulating,

00:01:44

or what content is being published using artificial intelligence.

00:01:48

The answer is not to ban everything.

00:01:50

A blanket ban would be impossible to enforce and

00:01:52

would push these uses even further into the shadows.

00:01:55

The right approach comes down to a few questions to ask before opening a tool.

00:02:00

Is it authorized?

00:02:01

What data am I going to give it?

00:02:03

Is this use sensitive?

00:02:05

Is human validation necessary?

00:02:07

Shadow AI is therefore not just a technical issue.

00:02:10

Above all,

00:02:11

it is a question of visibility and control.

00:02:14

In the next video,

00:02:15

we will move on to practice

00:02:17

with 3 simple habits for using AI responsibly.

Mandarine AI: WHAT YOU SHOULD KNOW

RELATED COURSES This tutorial is part of the following courses

AI Act : Mastering the European AI Regulation

E-learning Discovery

Vu 11 fois

AI Act : Mastering the European AI Regulation

Understand the EU AI Act, identify risky AI uses, and adopt best practices for responsible AI use in the workplace.

ALL COURSES