Why 48% of Your Employees Are Hiding Their AI Use From You | Human-Centered AI Leadership| AI for Executives - AI Cafe Conversations | Neuroscience, Neuroleadership, and Human-Centered AI for Executives

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_00 (00:13):
Welcome to the AI Cafe Conversations, your
favorite podcast.
This is the Friday Forbes LikeEdition.
I'm Sahari Nradi, your AIwhisperer.
I'm a neuroscience-basedneuroleadership consultant,
Forbes Coaches Council memberand influencer, and someone who
has been tracking the shadow AIconversation in organizations

(00:37):
for two years now.
I started talking about shadowAI even before people even knew
what it is.
And and to be honest, I stillhave people till today have
never heard the word shadow AI.
So let me give you a very umsimple definition of shadow AI.

(00:58):
Shadow AI means when people workin the shadow for with AI in
your organization, meaning thatthey are using tools and
platforms that are AI that arenot sanctioned by your
organization and they don't talkabout it.
That's what shadow AI is.
So let's talk today.

(01:19):
What I want to talk to you aboutis actually a number that
stopped me when I first saw it.
That number is 48%.

(01:43):
Basically, half of youremployees are hiding their AIUs
from you.
And that's according to researchfrom Henley Business School.
I'm not just making that numberup.
Okay, it's a real research.
Not using AI on occasion,actively hiding it.

(02:06):
And if you are an executivelistening to this, your first
instant might be that is acompliance issue.
That is a risk issue.
That is a policy issue.
And I give you all that.
All of that is correct.
But I'm going to offer you adifferent frame.

(02:27):
Because until you understand whythis is happening, no policy in
the world will stop it.
So this is the hidden truth.
Alright?
Why they are going undergroundusing AI.
Here is what the data actuallyshows about shadow AI.

(02:50):
AI tools used in the workplacewithout official approval.
The cyber news survey found that59% of employees overall use
shadow AI tools.
So one of them gives us 48%, theother one gives us 59%.

(03:13):
So it's half or a little bitover half of your team members
are using shadow AI underground.
The number jumps to 93% when youlook at executives and senior
managers.
Again, that's another number.
Almost all your executives andsenior managers, almost are

(03:38):
using shadow AI.
Which means that the people mostlikely to be using unapproved AI
tools are often the ones settingAI policy.
That's ironic, isn't it?
The UpGuard report surveyed 1500people across multiple countries

(03:59):
and found that 80% of workers oremployees use unapproved AI
tools.
50% do it regularly, and lessthan 20% use only approved
tools.
Now here is the piece that Ithink executives need to sit

(04:21):
with.
This is not because youremployees are reckless or
dishonest.
It has nothing to do with that,or they have a flaw in their
personality.
In most cases, it's because yourapproved tools are not meeting
their needs.
And I'm gonna share somethingvery personal.
It doesn't mean that it appliesto everybody, it doesn't mean

(04:43):
it's 100% correct, but this iswhat I have seen.
For example, most of theorganizations that I either hear
about or work with, they useMicrosoft Office.
So logically, they push theco-pilot, which is the AI branch

(05:05):
of Microsoft.
And as we are learning and as AIis developing, you cannot use
only one tool for all the needsthat you have, for all your
departments, for all their wantsand needs, and have one.
You have to create playbooks andworkflows with all of them.
And honestly, I personallyagain, this is a very personal

(05:27):
opinion.
Copilot never gave me everythingI ever wanted.
Again, that's a very personalopinion.
I'm not saying it's bad, it'sactually good, but for my use,
for what I do, it never gave mea complete response that lets me
feel like that's good, I'msatisfied now.

(05:48):
Okay, other people's experiencesare different.
I'm just sharing again, mine,very personal.
The cyber news data shows thatonly 33% of employees say the
tools their company providesactually do what they need them
to do.
So when you think about shadowAI as a compliance failure, you

(06:09):
are looking at the symptom.
Again, what do I always say?
Look at the root.
The root cause is a trust gap.
An eats gap.
AI underneath those, both ofthose, is a psychological safety
gap.
Employees are asking themselves,if I tell my manager I'm using

(06:33):
this tool to do my job better,what will happen?
Will I be penalized?
Will I be perceived as cuttingcorners?
Will I get in trouble for notusing the approved system?
When the answer to any of thosequestions is I don't feel safe,
being honest, they gounderground.

(06:55):
This is the brain in action.
This is what the brain isactually supposed to do or
created to do.
The amygdala, which is ourthreat center, processes threat.
Da.
When transparency feels unsafe,the threat response fires.

(07:16):
And the behavior that followshiding, concealment,
workarounds, is the nervoussystem doing its job, protecting
the person.
The question is, what is yourorganization doing that makes
transparency feel unsafe in thefirst place?

(07:36):
Now I want to discuss aboutsomething about the cost and
what is actually costing you,like I always say.
So let's talk about thefinancial reality of this
because this is where theconversation often shifts for
executives.
According to IBM research,shadow AI data breaches cost

(07:57):
organization$670,000.
$670,000 more than sanctioned AIbreaches.
More per breach.
And here is what it's importantto understand.

(08:19):
I actually call it the tax, theAI tax.
And here is what is important tounderstand.
Shadow AI breaches are not justmore expensive, they're more
common.
Because when people gounderground, they go outside
your security perimeter andguidelines.
They are using tools that do nothave your organizational

(08:39):
agreements, your datagovernance, your compliance
guardrails.
75% of shadow AI users sharesensitive information like
employee data, customer data,internal documents with
unapproved tools.
That is three-quarters of thepeople doing this.
Because a lot of people don'tknow how to put guardrails.

(09:27):
So this is something maybe youcan think about.
70% of employees are aware ofcolleagues doing exactly this.
The behavior is not hidden fromeach other, it's just hidden
from leadership.
Less than 50% of employees knowor understand what their
company's actual AI policy says.

(09:49):
So you may have a policy, but ifpeople do not know it, do not
understand it, or do not believeit applies to them, it does not
function as a policy.
It functions as a document.
And this is something that Iobserved working with so many
organizations.

(10:10):
There are more organizationsthat don't have an AI policy or
guideline or an MOU about how touse AI than those that have.
None.
There is no AI governancedocument that applies to all

(10:31):
employees of an organization.
Or very few.
It's rare.
Again, this is something that weneed to think about.
Here is the trust paradox I wantyou to see.
25% of workers say AI is theirmost trusted source of
information, on par with theirmanager, in some cases,

(10:53):
exceeding their manager.
Think about what that means.
Your employees trust an AI toolas much as they trust you or
more.
Not because the AI is moretrustworthy, but because the AI
does not judge them, does nothave a performance review

(11:16):
coming, does not have a reactionthey need to manage.
That is a human leadershipproblem, not AI problem.
The solution and how we closethe trust gap.
So what do you actually do withthis?
I'm going to give you threethings, as usual.
One mindset shift and twoconcrete actions.

(11:39):
One, the mindset shift stoptreating shadow AI as a
compliance problem and starttreating it as an intelligence
signal.
See the shift?
Shadow AI is your employeestelling you exactly what they
need that you're not providing.

(11:59):
It is the one of the most honestdata sets in your organizations
right now.
The question is whether you canhear it or not.
When I work with executives onthis, one of the frameworks I
use is called FAIR, F A I R.
It's specifically designed tohelp leaders make AI adjacent

(12:23):
decisions that are grounded inaccountability, impact, and
responsibility.
The first question it asks isnot who is breaking policy, it's
what is our behavior creating inthe people around us.
The question that changeseverything is that one.

(12:48):
Concrete action one.
A 90-day AI amnesty window.
No consequences, pureintelligence gathering.
Ask your teams what tools theyare using, what problems they
are solving, what gaps exist inyour approved systems.
You will learn more from thatconversation than from an adult.

(13:11):
The neuroscience here is verystraightforward.
Psychological safety, the feltsense that honestly will not get
you punished, activates theventral, the vagal system, the
one in our chest.
People in ventral vagal arecreative, transparent,

(13:31):
collaborative.
People in threat mode areprotective and concealed.
If you want transparency, youhave to create safety first.
Concrete action too.
Involve your teams in buildingthe AI policy, not just
communicating it, but buildingit.

(13:52):
The research is clear thatpeople support what they helped
create.
Always, we always said that evenin change management, to avoid
resistance or to lessen it.
When employees have a voice inthe boundaries, when they
understand the why behind theguardrails, compliance is not
about fear.

(14:13):
It's about ownership.
That is what MyCARES C-A-R-E-Sframework is designed for.
Creating the human conditionsfor people-centered AI adoption.
It's not about eliminating risk,it's about building the trust
infrastructure that makes riskmanageable.

(14:34):
And before I let you go, I wantto leave you with this.
Shadow AI is not the problem.
Shadow AI is the symptom.
The problem is that youremployees do not feel safe
enough to be honest with youabout how they are working.
And that's really sad.
And that is worth knowingbecause it's fixable.

(14:58):
If you want to know exactlywhere shadow AI is likely hiding
in your organization, I have afree five-minute assessment.
It's free.
That gives you a psychologicalsafety score and specific
recommendations link.
The link is actually here in thedescription.

(15:18):
And if you want to go deeper, ifyou want a conversation about
what this looks like in yourspecific organization, book and
no push, no sales, no pitch,leadership clarity call.
No pitch, just clarity.
The links are all below.

(15:39):
Leadership does not fail.
Nervous systems do.
And organizations thatunderstand this get ahead of the
problems that everyone else isstill reacting.
I am Sahar Andradi.
This is AI Cafe ConversationPodcast, the Forbes Like
Edition.
This is the only podcast whereneuroscience, AI, and leadership

(16:04):
interact together.
I will see you next Wednesday onour long podcast.

Before I go, I always say this: show me some love. (16:10):
undefined
Like, share, subscribe, comment,or leave us a comment on Google
or on Apple or any of thepodcast platforms that you see.
It helps us being distributed tomore people.
And thank you for making us oneof the top two percent global

(16:32):
podcasts.
Till we meet again.
Peace.

All Episodes

Why 48% of Your Employees Are Hiding Their AI Use From You | Human-Centered AI Leadership| AI for Executives

Episode Transcript

Popular Podcasts

Dateline NBC

Hey Jonas!

Betrayal Weekly

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Why 48% of Your Employees Are Hiding Their AI Use From You | Human-Centered AI Leadership| AI for Executives