New COLM paper introduces benchmark for evaluating how effectively AI models detect fabricated legal citations · Digg