The September 2026 infosec product roundup from Help Net Security shows a clear pattern: AI agents are now a first-class security concern. Several vendors announced controls that go beyond credential protection to govern what agents can do at runtime.
Ping Identity's Enterprise Personal Agent Access combines discovery, secretless access, and runtime control for personal AI agents. Akeyless's Agentic Runtime Authority enforces intent-based access control on top of its credential protection layer. Orchid Security addresses identity drift and provides application-level kill switches, noting that agents often exploit existing identity debt rather than breaking controls.
Other releases extend security to adjacent areas. Cohesity's Agent Resilience discovers, protects, and recovers AI agent infrastructure, while Akuity's Agentic Control Plane applies pipeline governance to AI agents shipping software. The roundup also includes non-agent products from Securin, BugBase, Bitsight, Scytale, and Superna, but the agent-focused releases stand out.
The source does not compare these products directly, so it is not possible to say which is best. Instead, the common theme is that enterprises are moving from asking what an agent can access to what it can do, and how to stop it if something goes wrong. That is a notable shift in how security teams approach AI adoption.