AI in safety-critical systems at DATE 2026

During the workshop held at DATE 2026 (April 20–22, 2026, Verona, Italy), industry and academic experts gathered to address a critical question: how to implement AI efficiently in safety-critical systems while guaranteeing reliability and certification.

The workshop tackled key challenges in embedded AI, including:

• The gap between training and implementation models
  
  
• Model specification and numerical accuracy
  
  
• Optimization under certification constraints
  
  
• Frameworks and standards with concrete, deployable solutions

Speakers included Pierre Gaillard (CEA-List), Filipo Perotto (ONERA), Eric Jenn (IRT AESE – Saint Exupéry), Dmitrii Kirov (Collins Aerospace), Dumitru Potop-Butucaru, Guillaume Iooss, Fernando Fernandes (Inria), and David Defour (Fondation Université de Perpignan Via Domitia).

This session brought together the embedded systems community to advance the state of the art in AI for safety-critical applications.
Overall, the workshop offered a practical overview of certification challenges and concrete directions for deploying reliable AI in industrial contexts.