A multi-agent AI system named 'Zealot' was developed to autonomously execute cloud penetration testing by chaining existing misconfigurations and known vulnerabilities. The system demonstrates AI's capability to accelerate exploitation of server-side request forgery (SSRF), metadata service credential theft, service account impersonation, and data exfiltration in cloud environments.
