What is Arcjet Agent Runtime Security: Observe, Enforce, Audit?

AI agents are software that can take real actions, such as reading and writing to databases, responding to support tickets or refunding payments, rather than only answering questions. Security leaders, Arcjet says, keep asking which agents are running, whether a given action should be allowed, and what happened afterwards. Arcjet Agent Runtime Security is a product from Arcjet for this problem; the maker describes Arcjet as an AI agent runtime security platform. It follows three steps named Observe, Enforce and Audit: discover which agents run in an organization, put an allow or deny decision in front of each tool call, and record what the agents did. An Arcjet blog post by David Mytton introducing it was first published on September 17, 2026.

Written from primary sources and reference pages, and checked against them by an independent AI fact-checker. Drafted with AI assistance.