1 articletagged with “auto-execution”
Attacks where AI agents are compromised without any user interaction -- processing malicious content triggers autonomous execution, data exfiltration, and system compromise.