Sweet Security Unfurls AI Agent to Conduct Penetration Testing
Security Boulevard, Wednesday, May 13th, 2026
Sweet Security, a provider of a cloud security platform, today added an artificial intelligence (AI) agent that makes use of an index of IT environments it maintains to conduct penetration testing.
Yigael Berger, chief AI officer for Sweet Security, said the Sweet Attack agent has been trained to provide a continuous red team testing capability that makes use of context, threat intelligence and semantic data to probe IT environments for weaknesses that could be exploited.
Based on the latest 5.5 AI model provided by OpenAI, the Sweet Attack accesses live, empirical, and indicative data trails that have been indexed to identify exploitable attack chains spanning multiple vulnerabilities, identity relationships, permissions, exposed application programming interfaces (APIs) and unauthenticated endpoints.