Neev Parikh

Parikh studies whether AI systems can hide harmful actions from oversight. He helped develop evaluations that test AI monitors^[1] and co-authored RE-Bench, which compares AI agents with human experts on machine-learning research and engineering tasks.^[2]