Bengüsu's work at Arq spans international AI governance coordination, scenario planning, and stakeholder engagement across European and global policy and tech audiences.
Previously, at the Centre for Future Generations, she co-authored Advanced AI: Possible Futures, Europe and the Geopolitics of AGI, and a portfolio of research for policy measures to increase societal resilience against AI risks covering frontier AI risk thresholds, open-weight models, and dual-use aspects. She has worked in the industry across tech and high-performance computing roles. She co-founded AI Safety Türkiye and is a published novelist.
Bengüsu holds an MA in Quantitative Methods in Social Sciences from Columbia University and a BSc in Industrial Engineering and Psychology from Sabancı University.
Publications
Arq publications
When AI stops thinking out loud
By Zoe Tzifa-Kratira, Fabio Marinello, Daan Juijn, Bengüsu Özcan
Today, AI models reason in language humans can understand, but more efficient architectures can lead to an AI-only language. There is a narrow policy window to preserve our ability to oversee AI models' behavior before AI developers give up on this over competitive pressure.
What we talk about when we talk about AI for science
By Bengüsu Özcan
"AI for science" is not a single category, and the distinction shapes what we should invest in, why, and how.
Preparing Europe for Transformative AI
By Arq Team
Arq Foundation is a new Brussels think tank helping Europe keep its agency, resilience, and flourishing in an age of transformative AI.
External publications
At Ankara, Nato has to admit a hard truth: its most vital AI infrastructure is built for business, not war
Argues that the AI infrastructure Nato increasingly depends on is commercial by design, and that the alliance must confront that dependency.
Europe and the geopolitics of AGI
The need for a preparedness plan