Responsible AI Platform
Critical Infrastructure

AI Act Compliance for Energy & Telecom

Smart grids, network management and critical infrastructure — high-risk under the AI Act

Practical guidelines for energy companies, network operators and telecom providers to comply with the EU AI Act.

View the compliance checklist

Why Take Action Now?

The AI Act has major impact on energy and telecom

August 2025

First obligations for high-risk AI systems in critical infrastructure come into effect

Critical Infrastructure = High-risk

AI in energy and telecom networks automatically falls under the strictest AI Act rules (Annex III)

Fines up to €35 million

Or 7% of global annual turnover — ACM and sector regulators will enforce

Safety Components

AI as safety component in networks requires conformity assessment and CE marking

High-risk AI in Energy & Telecom

These AI applications fall under strict AI Act requirements (Annex III)

Smart Grid Management

AI systems managing energy networks, distributing load and predicting grid congestion — essential for supply security.

Grid congestion predictionLoad balancing AIEnergy distribution optimizationOutage prediction

Predictive Maintenance Critical Infra

Systems predicting maintenance for power plants, transformer stations and telecom towers — outage can have societal impact.

Turbine monitoring AITransformer diagnosticsCable degradation detectionTower inspection AI

Network Monitoring & Cybersecurity

AI for detecting cyber threats and anomalies in energy and telecom networks — crucial for national security.

Anomaly detectionThreat intelligence AISCADA monitoringDDoS prevention

Energy Pricing & Customer Decisions

Algorithms determining energy prices, assessing customers or applying dynamic tariffs — direct impact on consumers.

Dynamic pricingCustomer profile scoringEnergy poverty detectionConsumption prediction

Specific Challenges for Energy & Telecom

The AI Act brings unique compliance questions for the sector

NIS2 and AI Act Overlap

How to combine NIS2 cybersecurity obligations with AI Act requirements? Where do they overlap and where do they conflict?

Real-time Decisions

Energy networks require millisecond decisions. How to comply with human oversight without compromising network stability?

Safety Component Classification

When is AI a safety component under the Machinery Regulation? And what does that mean for CE marking?

Legacy SCADA & OT Systems

Much operational technology is decades old. How to integrate AI Act compliance into existing industrial systems?

Supply Chain Complexity

Energy and telecom chains are complex. Who is provider, who is deployer in integrated AI solutions?

Sector-specific Supervision

ACM, RDI and State Supervision of Mines — how do these regulators interpret the AI Act for the sector?

AI Act Compliance Roadmap

Practical steps for energy and telecom companies

1

AI Inventory

2-4 weeks

Map all AI systems. Which systems manage critical infrastructure or make customer decisions?

2

Risk Classification

1-2 weeks

Determine per system if it is high-risk due to critical infrastructure, safety component or consumer decision.

3

Gap Analysis

3-6 weeks

Compare current documentation with AI Act, NIS2 and sector-specific requirements.

4

Remediation

3-12 months

Implement technical documentation, risk management, human oversight and cybersecurity measures.

5

Ongoing Monitoring

Ongoing

Set up processes for continuous monitoring, incident reporting and periodic review.

18-month trajectory

Implementation Roadmap

Detailed 6-phase trajectory with concrete deliverables

1

Inventory

Month 1-2
Complete AI system register (IT + OT)Owners per systemLink with NIS2 asset register
2

Classification

Month 2-3
High-risk vs. limited/minimal per systemSafety component assessmentCritical infrastructure mapping
3

Gap Analysis

Month 3-5
Gap per system: AI Act + NIS2 + sector legislationPrioritization based on risk and deadline
4

Governance Framework

Month 5-7
AI governance structureRoles & responsibilitiesIncident response procedure
5

Implementation

Month 7-14
Technical documentation per systemHuman oversight mechanismsCybersecurity integrationCE trajectories start
6

Audit-ready

Month 14-18
Internal auditDry-run for ACM/RDIContinuous monitoring operational

AI System Inventory

Typical AI systems in energy & telecom and their likely classification

Note: many OT systems contain AI without it being explicitly labeled as such. Also inventory embedded AI in SCADA, DCS and network management.

Grid & Network Management

Usually high-risk
Load balancingCongestion managementFrequency regulationNetwork topology optimization

Annex III — AI in critical infrastructure management is automatically high-risk

Predictive Maintenance

Context-dependent
Turbine monitoringCable diagnosticsTransformer analysisTower inspection

High-risk if failure poses safety risk; otherwise possibly limited risk

Cybersecurity AI

Context-dependent
Intrusion detectionAnomaly detectionThreat huntingSIEM analytics

High-risk for autonomous blocking; limited risk for alerting-only

Customer & Market

Often high-risk
Dynamic pricingCustomer scoringChurn predictionSmart meter analytics

High-risk when impacting access to essential services

Content Moderation (Telecom)

Limited risk
Spam filteringContent classificationNetwork traffic analysisFraud detection

Transparency obligations (Art. 50); DSA overlap for platforms

Internal Operations

Usually minimal risk
Workforce planningDocument processingInventory managementRoute optimization

Minimal risk unless it makes autonomous decisions about individuals

Classification Decision Tree

Quickly determine the risk classification of your AI system

Does the system manage critical infrastructure (energy grid, telecom network)?

Yes

Automatically high-risk

No

Go to next question

Is it a safety component in a machine or installation?

Yes

High-risk + CE marking required

No

Go to next question

Does it make autonomous decisions about consumers (pricing, access, disconnection)?

Yes

Probably high-risk

No

Go to next question

Is it supporting with human override and no safety impact?

Yes

Possibly limited risk

No

Consult an expert for classification

This is a simplified decision tree. Consult your legal team for definitive classification.

Governance Structure

Recommended organizational structure for AI governance in energy & telecom

Board of Directors / Management
AI & Data Governance Committee (IT + OT + Compliance)
AI System Owners per business unit
OT Security Team
Compliance & Regulatory Affairs
Internal Audit & Risk

Link AI governance to existing NIS2 compliance structure and asset management — don't build from scratch.

Key Roles

AI System Owner

Responsible per AI system for compliance and performance — both IT and OT

OT/IT Security Lead

Ensures cybersecurity of AI systems in operational environments

Human Oversight Officer

Oversight for high-risk systems — adapted for real-time environments

Regulatory Affairs Lead

Coordinates between AI Act, NIS2, Energy Act and sector supervision

Compliance Checklist for Energy & Telecom AI

Concrete checkpoints for each high-risk AI system

AI system registered in EU databaseArt. 49
Risk management system set up (incl. OT risks)Art. 9
Data governance & data quality ensuredArt. 10
Technical documentation complete (IT + OT)Art. 11
Logging & traceability configuredArt. 12
Transparency to users and consumersArt. 13
Human oversight arranged (adapted for real-time)Art. 14
Accuracy, robustness & cybersecurity testedArt. 15
NIS2 incident response integrated with AI ActNIS2 Art. 23
FRIA completed as deployerArt. 27
Conformity assessment completed (if safety component)Art. 43

This checklist applies per high-risk system. Combine with NIS2 compliance requirements for efficiency.

Common Mistakes

Avoid these pitfalls in AI Act implementation

Forgetting OT systems

SCADA, DCS and embedded AI in network equipment are often not included in the inventory.

NIS2 and AI Act as separate tracks

Both regulations overlap significantly. Integrate compliance tracks for efficiency.

Deeming human oversight impossible

Real-time systems require creative solutions: monitoring dashboards, anomaly alerts, escalation processes.

Not inventorying vendor AI

Network equipment often contains third-party AI. As deployer you are co-responsible.

Only involving IT department

OT engineers, network specialists and operations must participate from day one.

Underestimating smart meter data

Consumption data is personal data. AI analyses on smart meter data fall under GDPR and AI Act.

What Makes Energy & Telecom AI Different?

Sector-specific considerations

Critical Infrastructure Status

Energy and telecom AI falls under both AI Act and NIS2 cybersecurity directive

Real-time Requirements

Networks require uninterrupted AI decisions — human oversight must be implemented differently

Societal Impact

Energy or telecom outage affects millions of people — reliability requirements are extremely high

Converging Regulation

AI Act, NIS2, Machinery Regulation and sector legislation require an integrated compliance approach

Sector-specific regulation

Regulatory Overlap

How the AI Act connects with existing energy and telecom regulation

NIS2 Directive

Overlap: Cybersecurity, incident reporting, supply chain security

Practical tip: Combine AI Act monitoring with NIS2 security operations center (SOC)

Energy Act

Overlap: Supply security, consumer protection, smart meters

Practical tip: AI Act transparency requirements reinforce existing consumer protection

Telecommunications Act

Overlap: Network integrity, interception, content filtering

Practical tip: AI Act adds specific requirements for AI-driven content moderation

GDPR

Overlap: Smart meter data, customer profiling, automated decision-making

Practical tip: FRIA can partially overlap with DPIA — combine where possible

Machinery Regulation

Overlap: AI as safety component in installations

Practical tip: CE marking requires combined conformity assessment