igergov.com
← Writing
PublicationCritical infrastructureArtificial intelligence · 3 min · BG EN ES DE

What really helps when the city stops

A systematic review of how artificial intelligence helps restore energy, water, transport and hospitals, and where the evidence still falls short.

The problem

When power, water, transport or hospital systems stop, the losses grow with every hour: social, economic and environmental. That makes restoring critical infrastructure a question of resilience as well as of engineering. Artificial intelligence is more and more often put forward as a help in that restoration, but what we know about what actually works is scattered across research papers, laws, standards and technical documents.

What we did

We carried out a systematic review, following the PRISMA 2020 methodology, of the research literature since 2016 and of the official documents on the subject. Of 55 records found, 53 sources remained: 31 research papers and 22 official documents. The selection was entirely manual, with no automated tools or AI used for exclusion. The results are organised around four questions: regulation, restoration practice, supporting systems and families of AI models.

What it showed

  • Regulation shapes the architecture of restoration: NIS2, the Critical Entities Resilience Directive (CER), the AI Act and national measures set requirements for risk, reporting, continuity and the supply chain.
  • Practice relies on rehearsed continuity plans, coordination during cyber crises, prioritisation by dependencies and measurable recovery objectives.
  • Systems need to move from separate monitoring tools to connected platforms: SCADA, IoT sensors, GIS, decision support systems, simulations and digital twins.
  • AI adds value when it is constrained, explainable, validated and built into processes under human oversight. Many of the applications, however, have still only been tested in simulation or in a single sector.

Why then

The paper came out in May 2026, at the point when the new European rules on cybersecurity and on the resilience of critical entities are starting to apply in practice, and the AI Act sets requirements for high-risk systems. The review connects these rules to the real work of restoration and shows where AI can help right now.

What comes next

The practical continuation of this work is CRIMS, a crisis management platform for municipalities, now in public beta.

The paper

Ivo Gergov, Maksim Sharabov, Alexander Rusev, Georgi Tsochev. Critical Infrastructure Restoration and Artificial Intelligence Systems: Applications and Practical Limitations. Sustainability 18, 5297 (MDPI, 2026). doi:10.3390/su18115297