Anca Dragan
Heads AI Safety and Alignment at Google DeepMind; UC Berkeley associate professor who runs the InterACT Lab
- Heads AI Safety and Alignment (on leave from UC Berkeley), Google DeepMind1
- Associate Professor, EECS Department (on leave), University of California, Berkeley1
Anca Dragan is an Associate Professor in the EECS Department at UC Berkeley, currently on leave to head AI Safety and Alignment at Google DeepMind, where Dragan leads teams responsible for the safety of Gemini models. At Berkeley, Dragan runs the InterACT Lab, which works on algorithms for human-AI and human-robot interaction, including learning reward functions from human feedback. Dragan is a co-author of Google DeepMind's 2025 paper "An Approach to Technical AGI Safety and Security", which sets out technical approaches to misuse and misalignment risks.12
Work
An Approach to Technical AGI Safety and Security (external site: arxiv.org)Published research
Co-author of the 2025 paper, which identifies four areas of risk (misuse, misalignment, mistakes and structural risks) and focuses on technical approaches to misuse and misalignment.2
InterACT Lab (external site: people.eecs.berkeley.edu)Published research
Runs the UC Berkeley lab, whose work covers learning reward functions interactively from diverse human feedback and algorithms for human-AI collaboration and coordination.1
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.