In brief AI still struggles to understand negation, posing risks in critical domains like healthcare. An MIT study found that vision-language AI, in particular, can’t reliably grasp negative statements. Experts warn that AI’s…