Research
Featured research
Simple Features Beat Most AI Control Monitors
Shubhrangshu Debsarkar, Valmik Nahata
Technical researchOctober 30, 2026
Read researchTechnical research
Technical paper
Simple Features Beat Most AI Control Monitors
Shubhrangshu Debsarkar, Valmik Nahata
October 30, 2026
Technical paper
AgentCollabBench: Diagnosing When Good Agents Make Bad Collaborators
Aritra Mazumder, Shubhashis Roy Dipta, Shubhrangshu Debsarkar et al.
May 9, 2026

Technical paper
Identifying and Validating Emotion Concept Representations in Gemma 2 2B
Avery Li, Nia Mucher
August 13, 2026

Technical paper
A Direction for Some, Not All: Cross-Model Transfer of Steering Vectors and the Limits of Steering-Based Safety Interventions
Joshua Yoo
June 17, 2026

Technical paper
Geometric Concept Representations in Language Models: A 3D Hue-Saturation Manifold in Gemma 2 9B and Manifold Steering
Aarav Lodha
June 14, 2026
Policy briefs
Policy brief
1st place
The AI Safety Investigation Act: A Federal AI Incident Investigation and Reporting Authority
Logan Bradley, Ishan AjwaniApril 4, 2026
Policy brief
2nd place
The Making AI Governable for Americans Act (MAGA Act)
Owen Watzlavick, Adrian KlaitsApril 4, 2026
Policy brief
3rd place
AI Market Structure Reform Act
Ricardo Bruinton, Jason Chen, Ryan HealyApril 4, 2026
Fellowship capstone projects

Fellowship project
The Current Landscape for AI Evaluations and Where We Need to Land
Mustafa Lonandwala
May 14, 2026

Fellowship project
What Armenia’s Tech Emergence Can Teach Us About Compute Diplomacy
Hovsep Seferian
April 28, 2026

Fellowship project
Location Verification Is Not Enough: A Dual Framework to Combat Large-Scale AI Chip Smuggling
Seth Lifland, Shubhrangshu Debsarkar
April 24, 2026