Anca Dragan

Anca Dragan was born in Brăila, Romania, earned a BSc in computer science from Jacobs University Bremen in Germany, and then went to Carnegie Mellon UniversityCarnegie Mellon UniversityCarnegie Mellon University (CMU) is a private research university in Pittsburgh, Pennsylvania. It grew out of the Carnegie Technical Schools, founded with funding from Andrew Carnegie in 1900, which were renamed the Carnegie Institute of Technology in 1912 and took the present name after merging with the Mellon Institute in 1967. The university is known for computer science, robotics, engineering, and the arts, and has campuses in Qatar, Silicon Valley, and Rwanda. As of September 2025, it enrolled 16,582 students. Its current president is Farnam Jahanian. On September 30, 2026, the university announced a $3 billion gift from Ken Griffin and plans to build a new campus in Miami.Open the full entry for her PhD.⁠[1]

She pursued a PhD in robotics at Carnegie Mellon University from 2009 to 2015 under Siddhartha Srinivasa.⁠[2]

From 2015 she was an assistant professor of electrical engineering and computer sciences at UC Berkeley, becoming an associate professor in 2021; from 2017 she also consulted for Waymo.⁠[2]

At Berkeley she runs the InterACT Lab, which studies algorithms for human-robot and human-AI interaction, and she helped found the Berkeley AI Research (BAIR) Lab.⁠[3]

In October 2016 she, Dorsa Sadigh, S. Shankar Sastry and Sanjit A. Seshia presented a paper at IROS 2016 on a planning algorithm that lets an autonomous car infer a human driver’s style by probing their reactions with small actions such as braking or nudging in.⁠[4]

In August 2017, at 30 and as an assistant professor at UC Berkeley, she was named to MIT Technology Review’s 35 Innovators Under 35 in the Visionaries category; the profile said she works to distill complicated or vague human behavior into simple mathematical models that robots can understand.⁠[5]

In October 2017, discussing how self-driving cars deal with pedestrians and other vehicles, she described a car that will not act because any action could lead to a collision as “the freezing robot problem,” and she works on interpreting and predicting human behavior.⁠[6]

In 2018 she received a Sloan Fellowship, was selected for the IJCAI Early Career Spotlight and gave keynote talks at CoRL 2017 and IROS 2018; she had received the Okawa Foundation Award in 2017.⁠[7]

That March she, Minae Kwon and Sandy H. Huang presented “Expressing Robot Incapability” at HRI 2018, a study of how robots can use motion to express what they were trying to do and why they failed in tasks such as grasping.⁠[8]

In July 2019 she was one of nine UC Berkeley recipients of the Presidential Early Career Award for Scientists and Engineers (PECASE), nominated by the National Science Foundation, while an assistant professor and head of the InterACT Lab.⁠[9]

On March 19, 2020, she appeared as the guest of episode 81 of the Lex FridmanLex FridmanLex Fridman is a Soviet-born, US-raised computer scientist and podcast host who works as an MIT research scientist on human-AI interaction and autonomous driving. His Lex Fridman Podcast is known for hours-long conversations with guests including heads of state and technology leaders.Open the full entry Podcast, titled “Human-Robot Interaction and Reward Engineering,” whose outline covers the difficulty of human-robot interaction, humans in driving, semi-autonomous driving, how to specify good rewards and the three laws of robotics.⁠[10]

In April 2020 she explained on Marketplace that robots borrow economics’ approach of modeling human behavior as responding to incentives or utilities and use game theory to handle how people and robots influence each other, which she called “kind of a negotiation.”⁠[11]

In 2021 she received the IEEE Robotics and Automation Society (RAS) Early Academic Career Award in Robotics and Automation, having received the McEntyre Award for Excellence in Teaching in 2020.⁠[1]

In February 2024 Google DeepMind formed an AI Safety and Alignment organization, responsible for safeguards for Gemini models and for AGI safety research, with Dragan as its leader.⁠[12]

She said the team aims to help models better understand human preferences and values and resist adversarial attacks; on splitting her time with her Berkeley lab, she said the two lines of research are complementary.⁠[12]

On April 2, 2025, she, Rohin Shah, Four Flynn and Shane LeggShane LeggShane Legg is a New Zealand–born artificial intelligence researcher who co-founded DeepMind with Demis Hassabis in 2010 and is chief AGI scientist at Google DeepMind, with long-standing work on the theory of machine intelligence and the safety of artificial general intelligence.Open the full entry published a blog post introducing Google DeepMind’s paper “An Approach to Technical AGI Safety & Security,” which divides risks into misuse, misalignment, accidents and structural risks, with a deeper focus on misuse and misalignment.⁠[13]

As of October 2026 she is on leave from UC Berkeley, leading the teams responsible for the safety of Gemini models at Google DeepMind.⁠[3]

Personal website: Public link: https://people.eecs.berkeley.edu/~anca/ (opens in a new window)⁠[10]

X: Public account: https://x.com/ancadianadragan (opens in a new window)⁠[10]

UC Berkeley: Official website: https://www.berkeley.edu/ (opens in a new window)⁠[9]

Google DeepMind: Official website: https://deepmind.google/ (opens in a new window)⁠[13]