Google’s Threat Intelligence Group says it has “high confidence” a threat actor used an AI model to help discover and weaponize a vulnerability in a popular system admin tool.