Join Our Community

Valeriia is part of the organizing group of AI Safety for Italy, contributing to the coordination and development of the initiative across governance, finance, communications, and operational support. Her background is primarily in operations, project coordination, stakeholder relationship development, and fundraising across international initiatives and campaigns. She has experience supporting organizational processes, cross-functional coordination, and the development of operational infrastructure. Her recent involvement in AI safety includes participation in initiatives such as AI Safety Camp, Apart Research, and the Corda Democracy Fellowship.

Carola Caivano

Co-founder — Communications & Community

Carola is a co-founder of AI Safety for Italy, contributing to the development of the initiative in external communication, community strategy, and networking. Her background is in physics, and she currently works as a data scientist. Her involvement in AI safety includes participation in the ML4Good bootcamp and completion of the AI Alignment course by BlueDot Impact.

Elisabetta Rocchetti

Co-foununder — Operations & Community

Elisabetta is a co-founding collaborator of AI Safety for Italy and a PhD candidate in Computer Science at Università degli Studi di Milano, where her research focuses on mechanistic interpretability. Her work spans refusal behaviour, instruction-following mechanisms, training dynamics, and the encoding of aesthetic concepts, using methods such as activation patching, steering, probing, and complex network modeling.

Cristian Curaba

Technical Infrastructure & Didactics

Cristian holds a Master’s degree in Data Science and Scientific Computing and a Bachelor’s in Mathematics, providing him with a rigorous analytical and problem-solving foundation. He actively participated in the Apart Fellowship, ML4Good France, and various related courses. His involvement extends to local community building, where he currently leads an AGI Safety Discussion Group at the University of Udine to engage researchers and students in critical AI safety topics.

Francesco Ortu

Co-founder — Operations & Mentorship

Francesco is a co-founder of AI Safety for Italy and a PhD student at the University of Trieste and Area Science Park, where he studies the mechanistic interpretability of LLMs and VLMs. During his PhD, he spent some time as a research intern in the AI Safety and Alignment group at the ELLIS Institute in Tübingen.

Luca Scionis is a PhD student in the Italian National AI PhD Programme, affiliated with the University of Cagliari (sAIferLab). He is currently an ELLIS intern working on the AI safety of AI agents. His research spans adversarial machine learning, with a focus on the robustness of computer vision models, and AI safety, particularly the security of LLM-based agents and assistants.

Madhusudhan Pathak

Communications & Technical Infrastructure

Working on pioneering the Artificial Wisdom paradigm, a fundamental reconceptualization of AI alignment that positions meta-ethical reasoning and contextual understanding. This research challenges the conventional approach of treating ethics as post-hoc overlays on optimization systems, proposing instead that wisdom must be embedded at the foundational level of AI architecture.

Luca Mattiazzi

Website & Technical Infrastructure

An AI Engineer with several years of Software and ML Engineering experience and a biotechnology degree. Currently interested in mechanistic interpretability and generally in interpretable machine learning.

Lorenzo is a co-founder of AI Safety for Italy and a postdoctoral researcher at Area Science Park (Trieste). His research focuses on the interpretability of multimodal foundation models and representation alignment. He previously obtained a PhD in AI at the University of Trieste, supervised by Luca Bortolussi, with a research visit in Francesco Locatello’s Causal Learning and AI lab at IST Austria.