Arcjet launches runtime security for production AI agents

Arcjet launched agent runtime security on September 17, 2026. The product is designed for AI agents that take consequential actions across databases, support systems, payments, tools and APIs. It combines discovery of running agents with controls before and after actions, workflow observability and preserved audit evidence for security and compliance review. The release targets the growing gap between model-level safeguards and the behavior of multi-step agents operating in production systems. Source: Arcjet, September 17, 2026.