← Italian Observatory for AI
Vertical table 04 · Strategic infrastructures

AI in infrastructure Criticism

Energy, transport, water, healthcare, digital networks: where an algorithmic error does not remain confined to a single system but reaches essential territories and services. Here the Observatory studies how to deploy AI in these infrastructures in a reliable, secure and compliant manner.

LP
Luigi PinchiaroglioTable leadDesigner of large infrastructure projects, in Italy and abroad.
Why a dedicated area

More efficiency, but also new dependencies

Artificial intelligence is entering the operation and maintenance of networks, demand forecasting, emergency management, and plant protection. It improves the reliability and continuity of service, but links infrastructure to data, models, providers, and computing capabilities. In a critical infrastructure, a problem spreads.

Error or deviation from the modelA model that degrades over time or makes mistakes in a limited case can influence operational decisions regarding plants and networks.
Cyber attacksThe AI system can be a target, a defense tool or a new gateway.
Supply chainsData, models, the cloud, and external providers create dependencies that need to be known and governed.
ResponsibilitiesWho decides, who oversees and who answers when an automated system intervenes on an essential service.
The European picture

Three rules, one single interpretation

For those managing critical infrastructure, the AI system is both a regulated object, a digital component to protect, and a service resilience factor. Three disciplines view it from three different angles.

Regulation (EU) 2024/1689

AI Act

Classifies AI systems based on risk. Those used as components of digital network security, traffic, water, gas, heating, and electricity are high risk, with requirements on data, documentation, human oversight, and robustness.

In Italy: Law of September 23, 2025, no. 132
Directive (EU) 2022/2555

NIS2

It requests that cybersecurity of networks, information systems and providers be governed as a business risk, with responsibility at the highest level and notification of incidents.

In Italy: Legislative Decree no. 138 of September 4, 2024
Directive (EU) 2022/2557

CER

Look at the overall resilience of critical entities: prevent, resist, absorb, and restore the service in the face of natural, technical, intentional, or hybrid events.

In Italy: Legislative Decree of September 4, 2024, No. 134
It starts with the use case, not the standard

The Observatory reads every AI application using a single matrix that brings together compliance, cybersecurity, and service continuity, to avoid separate assessments and duplicate documentation.

The scope is completed with GDPR, Data Act, Cyber Resilience Act and the technical standards ISO/IEC 42001 and ISO/IEC 23894.

The work field

The sectors of essential services

The reference is to the sectors identified by the discipline on critical entities, supplemented by those of cybersecurity when analyzing suppliers and digital dependencies.

EnergyTransmission and distribution networks, power stations, gas pipelines, LNG terminals, district heating.
TransportRailways and signaling, airports and air traffic control, ports, highways.
WaterCollection, treatment and distribution; sewer networks, treatment plants and reuse.
HealthHospitals, emergency-emergency networks, laboratories, production of critical drugs and devices.
Digital infrastructureData centers and cloud, communication networks, Internet nodes, submarine cables, trust services.
Public administrationCentral digital platforms, demographic, fiscal and social security systems, digital identity.
SpaceGround control centers, satellite communications, Earth observation and navigation.
Food chainLarge processing plants, controlled-temperature logistics, cold chain.
The ENIA architecture

Six pillars applied to infrastructure

The table works within the vision of the Fondazione ENIA’s White Paper, which views technology, skills, governance, and participation as parts of a single public infrastructure for artificial intelligence.

1
Ontology and semanticsA common vocabulary for assets, events, risks and service levels, to enable data from operators, authorities and territories to be communicated.
2
Edge and sovereigntyWhere to process data and models: local inference when latency, continuity, and control are crucial.
3
AI LiteracyDifferent skills for leaders, designers, control rooms, procurement, CISO, DPO and maintainers.
4
AI governanceRoles and decision-making steps throughout the life cycle, from purchase to disposal.
5
Community JuriesTrust, risk perception and impacts on services, alongside technical indicators.
6
AI Finance and InsuranceDue diligence, insurance and the value of algorithmic assets linked to resilience.
An open table

Who we work with

The table is designed to bring together those who regulate, those who manage, those who build the technologies, and those who experience the effects of the services.

Institutions and authoritiesSectoral authorities and territorial bodiesOperators of infrastructure and essential servicesUniversities and research centresTechnological companies and the industrial chainAssociations, orders, finance and insuranceCivil society and territorial communities
How to participate

Your use case, within the Observatory

Operators and companies in the supply chain are implementing practical applications of AI. The Foundation is studying them with its own laboratories as a research activity. Some examples:

Predictive maintenanceModels that anticipate failures in networks, systems and lines.
Demand forecastEnergy loads, traffic flows, water consumption.
Sale control and emergenciesSupport for decisions, with human supervision and safe degraded modes.
OT SecurityDetection of abnormalities in industrial control systems.

Activities are carried out as Observatory research, not as consultancy. Terms and fees on request, set out in the partnership agreement.

Table Infrastructure Strategies

Let's talk about your infrastructure

Tell us about the AI application you are evaluating or already using: we will tell you how it can contribute to the Observatory’s work.