Anca Dragan

Also known as Anca D. Dragan

Anca Dragan is a computer scientist whose research connects human-robot interaction with AI alignment and safety. She earned a computer science degree at Jacobs University in 2009 and a robotics PhD at Carnegie Mellon in 2015, taught Berkeley’s Algorithmic Foundations of Human-Robot Interaction, and coauthored a cooperative model of value alignment. She received the Presidential Early Career Award in July 2019, worked as a Waymo research scientist in August 2019, and won an IEEE Robotics and Automation Society Early Career Award in 2021. She coauthored Google DeepMind’s Frontier Safety Framework in May 2024; the framework’s third iteration followed in September 2025, with an April 2026 update adding Tracked Capability Levels. As of October 2026, she is VP of AI Safety & Behavior at Google DeepMind and an associate professor in EECS at UC Berkeley.

Now at: University of California, Berkeley, Google DeepMind and Waymo
Works: 222
Citations: 10,478
h-index: 51
Badges: Rising
Bridge
Links: Website

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2009–2015

Computer science and robotics

She earns a B.S. in computer science at Jacobs University in Bremen in 2009 and completes a robotics PhD at Carnegie Mellon in 2015. Her thesis, “Legible Robot Motion Planning,” names Siddhartha Srinivasa as her advisor.
2015

Teaching human-robot interaction

In fall 2015, she teaches Berkeley’s Algorithmic Foundations of Human-Robot Interaction. The course asks how robots should move around people, learn from user feedback, and assist users with daily tasks.
2016

Cooperative inverse reinforcement learning

In 2016, she coauthors “Cooperative Inverse Reinforcement Learning” with Dylan Hadfield-Menell, Pieter Abbeel, and Stuart Russell. The paper models value alignment as a cooperative, partial-information game in which the robot learns a human’s reward function.
July 2019

A presidential early-career award

In July 2019, she receives the Presidential Early Career Award for Scientists and Engineers. She is an assistant professor in Berkeley EECS and heads the InterACT Lab.
August 2019

Research scientist at Waymo

In August 2019, she works as a research scientist at Waymo.
2021

An IEEE RAS award

In 2021, she wins the IEEE Robotics and Automation Society Early Career Award.
May 2024–April 2026

The Frontier Safety Framework

In May 2024, she coauthors Google DeepMind’s Frontier Safety Framework, which sets protocols to identify future AI capabilities that could cause severe harm and to detect or mitigate the risks. In September 2025, she coauthors the framework’s third iteration; an April 2026 update adds Tracked Capability Levels in certain domains to identify and evaluate emerging risks sooner.
As of October 2026

VP of AI Safety & Behavior

As of October 2026, she is VP of AI Safety & Behavior at Google DeepMind and an associate professor in UC Berkeley’s EECS Department. At DeepMind, she leads teams responsible for content, frontier, and AGI safety and alignment, as well as model behavior and character. Her Berkeley research uses interactive human feedback to help AI agents do what people want.

Career

Since 2025Google DeepMindVP of AI Safety & Behavior
Since 2015University of California, BerkeleyAssistant Professor
From 2019WaymoResearch Scientist
Department of Electrical Engineering and Computer Sciences, University of California, BerkeleyHead