Skip to the main content

Tom Zick is an AI researcher and attorney, currently serving as President of Trillium Labs, a nonprofit dedicated to building out the open research ecosystem for frontier AI. Her work focuses on open models, post-training methods and AI governance.

Tom has advised organizations ranging from early-stage startups to Fortune 500 companies on emerging AI and privacy issues, and has contributed to standards like the NIST AI RMF and AIUC-1. She has held roles in industry, building out the responsible AI team at Charles Schwab, in academia, where she partnered with leading frontier AI labs to build collaborations with Harvard’s research community around red teaming and model alignment, and in city government.

Tom holds a JD from Harvard Law School and a PhD in Astrophysics from UC Berkeley.

Projects & Tools

Responsible Generative AI: Accountable Technical Oversight

Generative AI is at a tipping point of adoption and impact, much like that of machine learning and AI years ago. But this time the stakes, regulatory environment, potential social…

News

May 31, 2023

Exploring the Impacts of Generative AI on the Future of Teaching and Learning

In December 2022, researchers from BKC, OpenAI, and Khan Academy and other experts gathered to discuss the impacts of generative AI on teaching and learning.

Community

arXiv

Comparing Apples to Oranges

A Taxonomy for Navigating the Global Landscape of AI Regulation

With Sacha Alanoca and Kevin Klyman, Affiliates Shira Gur-Arieh and Tom Zick “present a taxonomy to map the global landscape of AI regulation.”

May 19, 2025
EdTech Magazine

How Personhood Credentials Could Impact Higher Education

This new type of identity verification aims to separate humans from artificial intelligence.

Tom Zick weighs in on the utility of personhood credentials for combatting AI in higher education.

Nov 4, 2024

Personhood Credentials

Artificial intelligence and the value of privacy-preserving tools to distinguish who is real online

Wendy Seltzer and Tom Zick consider the efficacy of implementing "personhood credentials" as means to deter bad actors online while maintaining users' anonymity.

Aug 26, 2024
UC Berkeley Center for Long-Term Cybersecurity

Choices, Risks, and Reward Reports

Reinforcement learning could be the most promising path to artificial general intelligence. But Tom Zick and a team of other researchers urge caution in a new report.

Feb 1, 2022
BKC Medium Collection

Between Games and Apocalyptic Robots

Tom Zick considers the near-term societal risks of reinforcement learning

Apr 16, 2021