Usman Anwar is an AI safety researcher who did his PhD at the University of Cambridge's Computational and Biological Learning (CBL) lab, supervised by David Krueger. His research spans reinforcement learning, deep learning, and AI alignment, with a particular focus on chain-of-thought monitorability — developing methods to ensure AI reasoning processes remain transparent and auditable. He is a 2022 Open Phil AI Fellow and Vitalik Buterin Fellow on AI Safety. His PhD research has produced influential work including a landmark survey identifying 18 foundational challenges in assuring alignment and safety of large language models. His LinkedIn profile lists his current affiliation as Coefficient Giving, suggesting he has moved into a grantmaking or research role there after completing his PhD.
Funding Details
- Annual Budget
- -
- Monthly Burn Rate
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
- Fiscal Sponsor
- -
Theory of Change
Anwar's research focuses on making AI systems more interpretable and monitorable, particularly ensuring that chain-of-thought reasoning in large language models cannot be used to hide information from overseers. By developing formal frameworks and training objectives that prevent AI deception and steganography, he aims to make advanced AI systems safer to deploy and easier for humans to supervise. His broader work on alignment challenges provides a shared research agenda that helps the field prioritize and coordinate on the most important technical problems. Reducing the likelihood of deceptive, misaligned AI systems is a direct path to reducing existential risk from advanced AI.
Grants Received
from Open Philanthropy
Projects
No linked projects.
People
No linked people.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.
Details
- Last Updated
- Apr 2, 2026, 10:55 PM UTC
- Created
- Mar 20, 2026, 2:34 AM UTC