Sweet Security launches Virtual Patching for running cloud and AI workloads, tied to its autonomous AI red-team agent

Launched around Oct 30, Sweet's Virtual Patching applies runtime compensating controls to block exploitation of already-known vulnerabilities on live cloud and AI workloads before a permanent fix ships. It links to Sweet Attack, Sweet's autonomous AI red-team agent, which validates exploitable attack paths against live runtime context and can propose a virtual-patching policy for one-click enforcement; the protection stays attached as workloads scale or restart and is removable once patched.

It directly targets the exploit-before-patch gap that AI-accelerated vulnerability discovery widens (e.g., an Atlassian flaw exploited within hours of PoC), and connects AI red-teaming output directly to an enforced runtime control for production cloud/AI workloads.

Cloud runtime security and AppSec teams where patch deployment lags PoC exploitation; evaluate alongside remediation SLAs for workloads such as Node2Shell/HTTP2Shell.