A new analysis maps a year of real-world AI-enabled cyberattacks onto the MITRE ATT&CK framework to show how threat actors are weaponizing AI.