Autonomous AI Attack Chains

These terms describe how AI systems can be compromised and then used to power attacks that spread on their own, such as agentic ransomware or worms moving through multi-agent and multi-tenant AI infrastructure. Together they trace an escalation path: a single AI system is compromised, that compromise is weaponized into AI-enhanced attacks, and those attacks can turn into agents fighting for control of shared infrastructure.