Diogo Cruz

I'm a Technology Specialist at the European Commission's EU AI Office, based in Brussels. My background is in AI safety research and quantum computing, with a PhD in physics from Instituto Superior Técnico.

My research interests include evaluating autonomous AI systems, understanding their behavior during long-horizon tasks, and improving methods for assessing AI safety.

I've mentored AI safety research projects through SPAR, Algoverse, and AI Safety Camp. My research includes work on goal drift, reward hacking, long-context agent safety, multi-turn jailbreaks, unlearning, and neural network interpretability, with publications in TMLR and at workshops associated with NeurIPS, ICLR, COLM, and AAAI.