AI Value Gap and AI Act: Why 81% of Companies Fail to Turn AI into Real Business Value

Why are 82% of companies investing in artificial intelligence, yet 81% fail to generate real business value?

In this episode of Diritto al Digitale, hosted by Giulio Coraggio, technology and data lawyerr, we explore the growing AI value gap and its profound legal and strategic implications for businesses.

The episode provides a detailed legal analysis of why artificial intelligence initiatives often remain isolated and fail to scale, focusing on critical issues such as AI governance, AI Act compliance, and GDPR risks. It explains why treating AI as a simple technological tool—rather than a regulated system—creates both missed opportunities and significant exposure to liability.

We also examine the implications of high-risk AI systems under the AI Act, the regulatory challenges posed by AI agents and autonomous decision-making, and how evolving legal frameworks are shaping the future of AI adoption in Europe and globally.

A key focus is on governance: how companies can design structured AI governance models to enable scalability, ensure compliance, and turn AI into a true competitive advantage rather than a legal risk.

This episode is designed for legal professionals, in-house counsel, compliance officers, and business leaders dealing with AI regulation, data protection, and digital transformation.

Key topics include: AI value gap, AI governance frameworks, AI Act requirements, GDPR and automated decision-making, AI agents regulation, legal risk management, and scalable AI adoption.

If you are advising on or implementing AI strategies, this episode will help you understand why the real challenge is not the technology itself—but how it is governed, structured, and legally managed.

Send us Fan Mail

Episode Transcript

SPEAKER_00 (00:07):
There is a number that should make every boardroom
pause.
82% of companies are alreadyexperimenting with artificial
intelligence.
But 81% report no meaningfulimpact on their bottom line, no
marginal impact, no real impact.
There is not a technologyproblem.

(00:28):
It's a structure failure and itraises a fundamental legal and
strategic question.
Are companies truly deployingartificial intelligence?
Or are they exposing themselvesto risk without capturing value?

(00:51):
I'm Giulio Coraggio, atechnology and data lawyer at
the global law firm DLP.
This is the podcast DirtoDigitale, podcast where we
explore the intersection betweenlaw and innovation.
And today we are addressing acritical issue for businesses,
regulators, and legalpractitioners.
The AI value gap.
A gap between potential andreality, between investment and

(01:13):
return, and increasingly betweeninnovation and compliance.
Let's start from a point that isoften misunderstood, even at

(01:33):
senior management level.
Artificial intelligence is notjust another IT solution.
It's not a plug-and-play tool.
It's not simply a procurementdecision.
From a legal perspective, thismisconception is particularly
dangerous.
Because companies tend toapproach artificial intelligence
deployment as a technicalintegration, while the legal

(01:54):
framework treats it as arisk-based system requiring
classification, documentation,accountability, and continuous
monitoring.
Under the UAI Act, thisdistinction becomes critical.
An AI system is not a neutralinfrastructure.
It's a regulated object, and theway it is designed, trained, and

(02:15):
deployed determines the legalobligations attached to it.
What we are seeing is a mismatchbetween technological adoption
and organizational readiness.
On one side, companies areintroducing AI systems into
their operations.
On the other, they aremaintaining legacy processes,
governance structures, anddecision-making frameworks.

(02:37):
This creates legal frictionbecause AI introduces automated
decision-making elements,problemistic outputs, opaque
reasoning processes, andincreasingly autonomous actions.
From a regulatory standpoint,this translates into multiple
layers of exposure.
GDPR risks, particularly aroundtransparency, lawful basis, and

(02:59):
automated decision making.
AI Act obligations, especiallyfor high-risk systems, product
liability implications,contractual allocation of risk
in supply chains.
And yet, many organizations aredeploying AI without redesigning
their internal controls.
They are adding complexitywithout reducing legal
uncertainty.

(03:20):
So, why does this value gapexist?
Why do companies invest heavilyin AI but fail to generate
tangible returns?
The answer lies infragmentation.
In many organizations, AIinitiatives exist and some of
them work extremely well, butthey remain isolated,
uncoordinated, and legallyunstructured.

(03:42):
From a legal standpoint, this isa key issue because scaling AI
is not just a technicalchallenge, it's a compliance
challenge.
If a company cannot classify itsAI systems, assess their risk
level, ensure data governance,implement human oversight, and
document decision-makingprocesses, then scaling becomes
legally unsustainable.

(04:04):
And as a result, initiativesremain confined to pilot
projects.
They never reach enterprise-widedeployment.
There is another factor that israrely discussed openly.
Risk aversion driven by legaluncertainty.
Many organizations are notfailing to scale AI because they
lack capability.
They are hesitating because theylack legal clarity.

(04:26):
Concerns typically includeliability exposure for incorrect
or harmful outputs, regulatorysanctions under the GDPR or the
UAI Act, reputational riskslinked to bias or lack of
transparency, contractual riskswhen relying on third-party AI
providers.
And this leads to a paradox.
The more powerful the AI systemis, the greater the hesitation

(04:50):
to deploy it at scale becomes.
Legal uncertainty becomes abarrier to innovation.
This challenge becomes even morecomplex with the rise of AI
agents.
These are not traditional AIsystems.
They are systems capable ofplanning, making decisions,
executing actions, interactingwith external systems.

(05:10):
From a regulatory perspective,AI agents raise critical
questions.
Under the UAI Act, they qualifyas AI systems.
But their classification dependson how they are used.
And this creates a moving targetfor compliance.
Because the same agent can below risk in one context, high
risk in another.
Moreover, AI agents challengetraditional legal concepts.

(05:34):
Who is responsible for anautonomous action taken by an AI
agent?
How do you ensure humanoversight when decisions are
dynamic and continuous?
How do you document compliancewhen outputs are not fully
predictable?
These are not theoreticalquestions.
They are emerging legal risks.
Artificial intelligence is anextraordinarily powerful

(05:55):
instrument, but it is also adouble-edged sword.
When properly governed, it canenhance efficiency, improve
decision making, unlock newbusiness models.
When poorly managed, it canamplify errors, embed bias at
scale, create systemiccompliance failures, and trigger
significant liability.
The difference lies not in thetechnology, but in the

(06:18):
governance framework.
This brings us to the key point.
Governance is the missing link.
Without a structured AIgovernance model, companies
cannot prioritize use cases,assess legal risks, ensure
compliance or scale successfulinitiatives.
Effective AI governance requiresclear internal policies, defined

(06:38):
roles and responsibilities,integration between legal,
compliance, and technical teams,or risk classification,
frameworks aligned with the UAIAct, continuous monitoring and
audit mechanisms.
In other words, governancetransforms AI from
experimentation into strategy.
There is a common perception.
The regulation slows downinnovation.

(06:59):
But in the context of AI, thisis only partially true.
Well-designed legal frameworkssuch as the AI Act serve a
different function.
They provide common language forrisk classification, structured
compliance pathways, and aframework for building trust.
For companies that adopt aproactive approach, regulation
becomes a competitive advantage.

(07:20):
Because it enables scalabledeployment, investor confidence,
and a market credibility.
On the contrary, companies thatignore governance risk remaining
trapped in the value gap.
If there is one key takeawayfrom this discussion, it is
this.
The problem is not artificialintelligence, the problem is how
organizations are integratingit.

(07:42):
Technology is not thebottleneck.
Governance is, legal structureis, organizational
transformation is, and untilthese elements are addressed,
the value gap will persist.
So let me leave you with a fewquestions.
In your organization, is AItruly generating value, or is it
still confined to isolatedinitiatives?

(08:03):
Do you have a governance modelcapable of scaling AI while
ensuring compliance?
And most importantly, are youtreating artificial intelligence
as a strategic asset or as atechnological experiment?
If you found this episodeinsightful, feel free to reach
out to me at julio.com forcomments or suggestions.

(08:24):
And don't forget to subscribe tothe podcast, activate the
notification bell, and leave usa 5-star review on Apple
Podcasts or Spotify if youenjoyed this episode.
I'm Giulio Coraggio, this is thepodcast Dirital Digitale.