Sotabase

Career

· Founder, Oasis Labs2018–
· Co-founder, Menlo Security2013–
· Founder, Ensighta Security2011–2012
· Professor, Computer Science Division, University of California, Berkeley2007–
· Assistant Professor, Carnegie Mellon University2002–2007
· PhD in Computer Science, University of California, Berkeley1999–2002
· M.S., Computer Science, Carnegie Mellon University1997–1999
· B.S., Physics, Tsinghua University1992–1996

Publications (28)

Representation Engineering: A Top-Down Approach to AI Transparency
arXiv.org · 2023
739
cited
Managing extreme AI risks amid rapid progress
Science · 2023
339
cited
The False Promise of Imitating Proprietary LLMs
arXiv.org · 2023
256
cited
On the Societal Impact of Open Foundation Models
International Conference on Machine Learning · 2024
86
cited
Formal Mathematical Reasoning: A New Frontier in AI
arXiv.org · 2024
72
cited
AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policies
AGI - Artificial General Intelligence - Robotics - Safety & Alignment · 2024
51
cited
Joint Language Semantic and Structure Embedding for Knowledge Graph Completion
International Conference on Computational Linguistics · 2022
43
cited
Forecasting Future World Events with Neural Networks
Neural Information Processing Systems · 2022
38
cited
Progent: Programmable Privilege Control for LLM Agents
arXiv.org · 2025
34
cited
Type-Constrained Code Generation with Language Models
Proc. ACM Program. Lang. · 2025
26
cited
How Would The Viewer Feel? Estimating Wellbeing From Video Scenarios
Neural Information Processing Systems · 2022
11
cited
Out-of-Distribution Detection Methods Answer the Wrong Questions
arXiv.org · 2025
9
cited
Advancing Science- and Evidence-based AI Policy
Science · 2025
8
cited
VERINA: Benchmarking Verifiable Code Generation
arXiv.org · 2025
7
cited
Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation
arXiv.org · 2025
5
cited
Position: Formal Mathematical Reasoning - A New Frontier in AI
International Conference on Machine Learning · 2025
3
cited
Hypervisors as a Foothold for Personal Computer Security: An Agenda for the Research Community
2012
1
cited
. Paper Number : 284 Title Page A Study of the Structure of the Web
1
cited
Where Confabulation Lives: Latent Feature Discovery in LLMs
Conference on Empirical Methods in Natural Language Processing · 2025
1
cited
FrontierCS: Evolving Challenges for Evolving Intelligence
arXiv.org · 2025
Sotabase