Sweet Security Launches Sweet Attack: Continuous Agentic Red Teaming Against Live Production Environments

SecurityWeek·medium signal

Sweet Security shipped Sweet Attack (GA) on May 13 — an AI red-team agent that uses runtime intelligence already indexed by Sweet (app behavior, identity paths, exposed interfaces, deployed source code) to discover and validate exploitable attack chains in live production. Unlike traditional penetration testing, Sweet Attack validates each chain step-by-step and drops unexercisable routes, logging validated chains with evidence, exploitation steps, and remediation paths. The company frames the launch around the 'Mythos Moment' — the realization that human security teams cannot match AI-assisted attack speed.