Measuring AI capabilities and risks.

Chris is the CEO of Measuring AI Progress and CEO and co-founder of EquiStamp, which he started in 2023 to provide engineering support for AI safety research. Under his leadership, EquiStamp has developed research infrastructure and benchmarks for leading organizations in the field, including HCAST for METR, ControlArena for the UK AI Safety Institute, and LinuxArena for Redwood Research.

He holds a BSc in Computer Engineering from Northeastern University and previously worked on machine learning at Amazon.

Daniel is an assistant professor of Computer Science at the University of Illinois, where his research focuses on rigorous measurement of AI progress and safe deployment of AI systems. His lab created the AI agent benchmarks CVE-Bench and InjecAgent, and he contributed to the widely used ML benchmarks DawnBench and MLPerf. He provided external safety review for OpenAI's gpt-oss preparedness evaluations and has advised Google, Microsoft, and a range of startups and funds.