Sweet Security Launches AI-Driven Sweet Attack to Operationalize Mythos-Grade Defense

New updates have been reported about Sweet Security.

Sweet Security has introduced Sweet Attack, an AI red-team agent that runs directly against customers’ live production environments, positioning the company at the center of enterprise efforts to answer Anthropic’s Mythos benchmark and AI-accelerated threats. Built on Sweet’s execution-level runtime data — including topology, Layer 7 exposure, deployed source code, identity paths, and real application behavior — the product maps, tests, and confirms exploitable attack chains at machine speed, providing CISOs with concrete evidence of real breach paths rather than theoretical risk scores.

The launch moves Sweet Security beyond detection into full-cycle attack-path validation and prioritized remediation, with early customers reporting that Sweet Attack uncovered exploitable routes missed by tier-one offensive security firms and traditional vulnerability workflows. By safely executing attacks and abandoning non-viable paths, the tool shifts security teams from broad vulnerability patching to focused breach prevention, giving boards and executives a quantifiable way to measure and close the Mythos-defined gap between attacker and defender capabilities. This new offering, now generally available to all Sweet customers, deepens Sweet Security’s role as a core platform for dynamic cloud and AI environments and may strengthen its competitive positioning and attractiveness to investors backing AI-native security platforms.