The Alice blog has a condensed version of an essay its biosecurity specialist wrote for AI Frontiers. The essay is about who researchers test on when they check whether AI helps with dangerous biological work.