AWS deployed AWS Security Agent, an AI agent that automatically reviews application code and architecture for vulnerabilities and runs penetration tests on demand throughout the software development process. It launched in preview in December 2025 and the penetration testing capability reached general availability in six regions on March 31, 2026.

AWS Security Agent takes application architecture documents, source code, and organizational security policy definitions as input and produces security violation reports, vulnerability findings with CVSS risk scores, attack chain documentation, and remediation suggestions as output. It covers three phases: design review (analyzing architecture documents against security requirements), code review (scanning code against organizational standards integrated into pull request workflows), and on-demand penetration testing (autonomously executing multi-step attack scenarios against running applications). At launch, the design review and code review features were available as preview in US East (N. Virginia); the penetration testing feature reached general availability in six regions (N. Virginia, Oregon, Ireland, Frankfurt, Sydney, and Tokyo) on March 31, 2026. AWS Security Incident Response, a related service, adds AI-powered investigation capabilities to accelerate security event response.