Research from Threatdown highlights that cybercriminals are weaponizing frontier AI models like Grok to commit cybercrime.