Threat actors are expanding their use of multi-agent AI attack frameworks, according to researchers at Google’s Threat Intelligence Group (GTIG).