Behavioural evaluation of AI systems
How models behave in context, under pressure, and across languages and cultures, including Italian and other under-represented settings.
We study AI systems in context: how they behave under pressure, how they fail across languages and cultures, and how technical evidence can support public-interest governance.
Research areas
Multi-agent and compositional safety
How individually aligned components can create emergent dynamics once models, tools, and agents interact.
AI, fundamental rights, and governance
How technical properties of AI systems interact with legal and political structures designed to protect people and communities.
Language, interpretation, and the humanities
How linguistic, philosophical, and cultural analysis reveals model behaviours that narrow benchmarks often miss.
Safety in deployment
How safety properties change when systems move from laboratory settings into institutional and commercial pipelines.
Interdisciplinary methodology
Shared protocols that let computer science, law, linguistics, and philosophy describe AI behaviour together.
Method
How we work
Icaro Lab works with external researchers, universities, and institutions in Italy and abroad. We publish preprints, build open methodologies, and translate technical findings into evidence that can support public-interest governance. Within our statutory purposes as a Third Sector foundation (Fondazione Icaro ETS), we also carry out applied evaluations and analyses in cooperation with public bodies and institutions of general interest — framed as research applied to the protection of rights and the safety of deployed systems, not as commercial services.