From 16 to 18 June 2025, the European Commission held the AI for Preparedness workshop, gathering experts from public institutions, research, industry, and civil protection to discuss how artificial intelligence (AI) can significantly enhance disaster risk management (DRM) in line with the EU’s Preparedness Union Strategy (PUS).
Day 1: Strategic foundations
Opening the event, Deputy Director-General of DG ECHO Hans Das outlined the urgent context: more frequent and severe natural hazards, new security challenges, and limited financial resources. He stated that: “AI has the potential to improve the efficiency and effectiveness of disaster prevention, early warning, and response systems.” While acknowledging that AI is not yet part of the modus operandi of DRM, he also highlighted the need to bridge the gap between research and operational use through tools such as pre-commercial procurement.
New technologies such as Artificial Intelligence can be game changing, but we need to invest in closing the gap between funding research and actually getting the projects into operation
Two panel discussions looked at how AI is being developed and used for DRM in Europe. Speakers shared examples such as digital twins, early warning tools, and research projects designed with input from civil protection authorities. They also stressed the need for clear data, ethical standards, and practical tools. A second panel focused on turning innovation into action: Horizon Europe projects were introduced as examples of ethical, user-friendly AI developed closely with practitioners.
Private companies ICEYE and OroraTech showed how satellite-based AI is already helping monitor floods and wildfires in real time. The panels underlined the importance of public-private partnerships and making AI tools usable in real emergencies.
A recurring message across the day was the importance of trust, usability, and collaboration. While the technological potential is clear, participants expressed the need for regulatory clarity, inclusive design, and end-user engagement to ensure AI delivers practical value in disaster settings. A science for policy brief on AI approaches to DRM was also launched on the first day, marking an important step in gathering knowledge of the array of AI-based tools already being used at the Joint Research Centre.
Day 2: AI in action
The second day of the workshop shifted from strategy to application, offering a deep dive into how prototype AI-based instruments are being developed to support DRM. Through a series of technical sessions, participants explored practical tools, data platforms, and robotics systems developed under EU initiatives and national programmes.
The Destination Earth initiative showcased two digital twins used for climate adaptation and short-term disaster forecasting like floods and heatwaves.
Another session by the Joint Research Centre (JRC) focused on explainable AI, emphasising transparency and easy understanding of AI outputs.
AI is not a standalone solution. It must be embedded into operational workflows and linked to legal, ethical, and societal safeguards.
Several EU-funded and Member State initiatives were also presented:
Throughout the day, a consistent theme emerged: the importance of user-centred design, interoperability, and ethical implementation. While the tools presented showed potential, were technically advanced and grounded in real-world needs—still significant cross-collaboration and cooperation is needed to prove that AI-based tools can enter operational use.
Day 3: Embedding AI into DRM systems
The third and final day of the AI for Preparedness workshop focused on shaping the path forward. While earlier sessions explored existing technologies, Day 3 shifted attention to practical challenges, procurement tools, and capacity-building needs for AI uptake across the civil protection community.
Breakout groups identified key areas where AI helps the most: early warning, real-time awareness, and quickly assessing impacts. One participant noted, “We are losing knowledge every time a report goes unread. AI could help us extract, summarise, and retain it.”
Challenges identified included fragmented data, trust issues, and the need for better infrastructure. Groups suggested creating clear, simple training programmes focused on practical use, ethical considerations, and tailored to different user needs. Participants also requested clearer guidelines for buying AI tools, stronger cooperation between different ministries, and more involvement from users in creating AI solutions.
Training, trust, and tailoring to end users—that’s what will determine whether AI becomes operational.
In closing, DG ECHO acknowledged these insights, stressing a collective need for investment in training, coordination, and governance frameworks. The workshop reinforced the EU’s commitment to integrating innovation into disaster preparedness and response. Feedback from these sessions will guide the next steps under the Preparedness Union Strategy.
Next concrete steps include: