With the rapid growth of artificial intelligence (AI), many professionals are exploring whether its power and capabilities could be an asset in managing emergency and crisis situations, such as extreme weather events and pandemics. At the same time, experts warn of the potential challenges of using AI in such complex, high-pressure, demanding environments. Is it possible to use AI responsibly, ethically and safely? Are we able to safeguard fundamental liberties? How far can we rely on AI in situations that could be dangerous and even life-threatening? Is it possible to build trust in AI? Drawing on the recent work for the European Scientific Advice Mechanism, our distinguished panel of speakers, discussed these issues, drawing on their academic and professional expertise and experience.
Rewatch this interactive webinar, which was free and open to all.
The conversation brought together researchers, legal experts, ethicists, and a firefighter, and the answers were far from straightforward.
Prof. Tina Comes described a "Goldilocks dilemma": too little trust in AI means useful tools go unused, but too much trust can lead to dangerous autopilot behaviour, especially under the stress of an emergency. The challenge is finding the right balance, depending on the type of AI and the stakes involved.
Prof. Andrej Zwitter walked through the EU AI Act and showed that while the regulatory framework provides some clarity, significant grey zones remain, particularly around accountability and the fact that most large AI models are developed outside Europe.
Dr. Olya Kudina drew on the COVID-19 pandemic to illustrate how a crisis accelerates AI adoption, often at the expense of proper validation. Over 700 diagnostic AI models were developed in early 2020 alone, many without robust testing. Contact tracing apps failed not because the technology was flawed, but because governments communicated poorly, and public trust eroded.
Commandant Quentin Brot offered a practitioner's perspective from the French firefighting services: AI works well for wildfire detection and logistics planning, but operational chatbots that advise firefighters during emergencies are, in his words, "highly dangerous." Nothing replaces human judgment, empathy, and courage on the ground.
Prof. Rémy Slama, speaking for the Group of Chief Scientific Advisors (GCSA) of the European Commission, argued for sovereign European AI tools and urged caution: AI should not currently be used for any decision in a crisis. Humans must remain at the centre. And investment in crisis prevention and preparedness is essential.
What made this webinar particularly valuable was the diversity of perspectives, from law to ethics, from academic research to firefighting on the ground, all converging on one message: the question is not whether AI has potential in crisis management, but whether we have the governance, the data quality, and the institutional readiness to use it responsibly.