Topic starter
02/06/2026 6:59 pm
TL;DR: AI agents are expanding enterprise identity attack surfaces because they authenticate, call tools, and act across cloud and SaaS systems with privileges that teams often cannot fully inventory or govern, according to Silverfort. Runtime identity controls at the gateway layer shift enforcement earlier, but they also expose how incomplete most agent governance models still are.
Editorial analysis by NHI Mgmt Group, based on content published by Silverfort: “Silverfort Secures AI Agents on Google Cloud in Runtime with Agent Gateway Integration”.
Key questions
Q: What breaks when AI agents are never deprovisioned?
A: When agents are never deprovisioned, they become zombie identities that continue to consume resources and preserve access long after their business purpose ends.
Q: Why do over-privileged AI systems increase lateral movement risk in cloud environments?
A: Over-privileged AI systems expand the blast radius of any mistake, compromise, or bad automation decision.
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
- Map every AI agent to a human owner Maintain a live register that links each agent to the person or team responsible for its scope, approvals, and exceptions.
- Enforce gateway-layer authorization Require policy checks at the point where agent requests enter downstream systems so overreach can be blocked before tool calls complete.
- Separate agent scope from human access Do not assume the permissions granted to the user who created an agent should automatically flow into the agent's runtime path.
Bottom line: AI agents create identity risk because they can authenticate, call tools, and take actions across multiple systems faster than static governance processes can track.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
02/06/2026 7:14 pm
Runtime identity control becomes the decisive boundary for agentic access. AI agents are not governed well by provisioning-time checks alone because their value and their risk emerge during execution. When an agent can choose tools, traverse services, and change context within a workflow, the control point has to move to the moment of action. Practitioners should read this as a governance shift, not a feature update.
A few things that frame the scale:
- Only 44% of organisations have implemented any policies to manage their AI agents, despite 92% agreeing that governing AI agents is critical to enterprise security, according to the 2026 Infrastructure Identity Survey.
- 69% of security leaders agree identity management must fundamentally shift to address agentic AI systems, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: What should organisations do when an AI agent has more privilege than its workflow needs?
A: They should reduce the agent's reachable systems, narrow the tools it can invoke, and separate the agent's runtime permissions from the human user's broader rights. The goal is to limit blast radius so a single compromised workflow cannot expand into unrelated data or services.
👉 Read our full editorial: Runtime identity security for AI agents in Gemini Enterprise