Topic starter
22/07/2026 4:35 pm
TL;DR: Agentic AI systems now chain tools, call APIs, and act without human approval at every step, exposing access paths that traditional controls were never built to contain, according to Appgate. The core issue is not model safety alone, but whether identity, privilege, and network reachability are enforced tightly enough to prevent autonomous misuse.
Editorial analysis by NHI Mgmt Group, based on content published by Appgate: “Impossible, Not Tedious: How AppGate ZTNA Meets Anthropic's Zero Trust Standard for AI Agents”.
Key questions
Q: What breaks when AI agents are given broad standing access?
A: Broad standing access breaks governance because the agent can move from one task to another without a fresh authorization check.
Q: Why do autonomous agents make traditional access reviews less effective?
A: Access reviews assume permissions persist long enough to be observed, challenged, and recertified.
Q: How can security teams tell whether agent access is actually under control?
A: Look for evidence that the team can trace every tool call, secret use, and cross-system action back to a named owner and a valid approval path.
Practitioner guidance
- Define agent trust boundaries before deployment Document which services, APIs, and data stores each agent is allowed to reach, and make that set smaller than the default platform reachability.
- Replace static shared credentials with verifiable machine identity Use short-lived, identity-bound authentication so each agent instance is authenticated and authorized separately instead of relying on reusable secrets.
- Segment agent workloads from adjacent systems Apply workload-level isolation so compromise of one agent does not create an easy path to internal pivoting or cross-service discovery.
Bottom line: Agentic AI turns access reachability into the main governance problem, because autonomous systems can act across APIs and internal services without human pacing.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
22/07/2026 4:40 pm
Zero Trust for agentic AI is fundamentally an access architecture problem, not a model-safety problem. The article is right to separate prompt defenses and governance overlays from the question of whether the system is reachable at all. Once an autonomous system can reach APIs, internal services, and orchestration layers, the security outcome is determined by policy enforcement at the connection point. Practitioner conclusion: if the agent can reach too much, every other control is compensating for a design flaw.
A few things that frame the scale:
- 69% of security leaders agree identity management must fundamentally shift to address agentic AI systems, according to the 2026 Infrastructure Identity Survey.
- 53% of security leaders expect AI to run major portions of their infrastructure autonomously within the next three years, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: Should organisations use zero trust or broad segmentation for AI agents?
A: Zero trust is the more defensible model because it enforces explicit policy at each connection point and narrows what the agent can reach. Broad segmentation can still leave implicit trust inside the environment, which is exactly where autonomous misuse turns into lateral movement and blast radius.
👉 Read our full editorial: Zero trust access controls are now foundational for agentic AI