On 22 October 2024, the European Commission (DG ECHO) held a landmark workshop under the Union Civil Protection Mechanism (UCPM) the Knowledge Series to explore how Artificial Intelligence (AI) and Machine Learning (ML) can transform Disaster Risk Management (DRM) across Europe.
Erwan Marteil, Head of Unit DG ECHO.B3, opened the event, emphasising the importance of AI as a “force multiplier” for enhancing resilience and preparedness against natural hazards. He underscored that while AI and ML hold tremendous potential for initiatives such as early warning systems and climate impact forecasting, they have its own pitfalls and are not yet systematically incorporated into DRM processes. The event’s aim was twofold: to present current EU-driven AI initiatives relevant to DRM and to collaboratively envision the steps needed to integrate these tools into the sector in an effective way.
Part of the workshop, Martin Bailey from DG CNECT highlighted the European AI Office's work on implementing the AI Act, which focuses on developing trustworthy AI frameworks. Andrea Toreti from JRC presented efforts within the UN Focus Group on AI for Natural Disaster Management, detailing collaborative initiatives like a standardization roadmap for AI in disaster management. Stephan Siemen from the European Centre for Medium-Range Weather Forecasts (ECMWF) introduced the European Commission flagship initiative Destination Earth, a major project developing a digital model of the Earth to simulate natural hazards and related human impacts.
Several real-world applications of AI in DRM were shared by speakers from both the public and academic sectors. Michele Ronco from JRC outlined the impact of AI on drought risk assessments, food crisis prediction, and wildfire response systems. Elena Xoplaki from Justus Liebig University Giessen discussed the MedEWSA project, which uses AI to reduce local communities’ vulnerabilities to hazards, creating forecasting and impact assessment tools. Francesco Focacci from Università eCampus and Francesco Pistolesi from the University of Pisa introduced the MEDEA project, which combines AI systems to estimate earthquake-related damages and psychological impacts, providing essential tools for prevention and post-disaster response.
Following the panels, participants divided into breakout groups to discuss questions on advancing AI in DRM. These sessions yielded insightful recommendations on capacity building, data accessibility, and the development of explainable AI outputs. Participants underscored the need for targeted training, open resources, and EU-developed learning materials to build widespread understanding of AI’s principles, risks, and practical applications across all levels of DRM.
The discussions also highlighted key challenges that must be addressed. In particularly, data gaps hinder accurate forecasting and situational awareness. Furthermore, transparency is critical; without it, citizens may struggle to understand or trust AI-driven insights. This trust issue is compounded by the lack of explainability in many AI systems—if citizens can’t see or understand the risks themselves, they may be reluctant to rely on AI-generated assessments. The need for transparency and trustworthiness in AI-driven insights is therefore critical in building public trust and ensuring effective disaster preparedness.
Additionally, participants highlighted that while AI outputs hold promise, it’s essential to address potential biases, quantify uncertainties, and ensure compliance with the AI Act. In particular, careful attention to data use is needed in areas concerning psychological vulnerabilities and economic impacts. They also stressed the need for sustained investment in computing resources, especially to support smaller players and organizations with limited capacity.
As a first step towards advancing AI in DRM, the workshop facilitated the establishment of an "AI for DRM" Knowledge Page on the UCPM Knowledge Network website as part of the Knowledge Series, providing open resources and existing training materials.
Participants agreed that further engagement will be essential to building a robust framework for AI-driven disaster management. Future plans may include workshops and events to foster collaboration among stakeholders, both within and outside the Commission, as well as a potential high-level event in 2025 aligned with the new Global Initiative on Resilience to Natural Hazards through AI Solutions.
In sum, the workshop highlighted that, with the right investments, partnerships, ethical considerations, and citizen engagement AI can be a powerful ally in Europe’s disaster resilience efforts. Through continued dialogue and development, DG ECHO and its partners are paving the way for an innovative and more resilient future in DRM.
Sectors
Risk drivers
Thematic series