TL;DR: Approov details that less than 10% of total traffic can already reshape security priorities when it is agentic and influential. Identity controls now have to verify non-human action as well as origin before access reaches production systems.
Editorial analysis by NHI Mgmt Group, based on content published by Approov: “RSAC 2026: From Hype to Agency”.
By the numbers:
- agentic traffic is still less than 10% of total volume
Key questions
Q: Why do least-privilege models need to change for agentic AI?
A: Least privilege still matters, but agentic AI changes the timing and granularity of enforcement.
Q: How should security teams verify identity when AI-generated profiles look authentic?
A: Security teams should stop relying on visual realism or message quality as proof of identity.
Q: How should security teams govern AI agents that can access enterprise systems?
A: Security teams should govern AI agents as non-human identities with explicit ownership, scoped privileges, and continuous monitoring.
Practitioner guidance
- Separate agent governance from generic bot monitoring Treat AI agents, automation and ordinary bots as different control classes.
- Bind authorisation to each action request Move policy decisions closer to the transaction so that an agent, workload or user is checked against purpose, scope and session context before the action runs.
- Inventory where cryptographic proof is still missing Map the systems that rely on channel trust, static tokens or weak origin checks and identify where stronger proof of entity and intent is required.
Bottom line: Agentic traffic is forcing security teams to govern behaviour at the action level, not just authenticate the source of a request.
Glossary terms: Agentic Traffic, Resource-Level Authorisation, Cryptographic Proof, Task-Scoped Least Privilege
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
Agentic traffic creates an identity governance problem, not just a detection problem. The article’s strongest implication is that security teams can no longer treat bots, workloads and agents as a single behavioural class. A system that can take actions autonomously or semi-autonomously needs governance that understands who or what is acting, under what authority, and for which task. That is a different control question from traditional perimeter filtering, and it is why programme owners should separate identity assurance from traffic classification.
A few things that frame the scale:
- Organisations that describe themselves as confident in their AI deployment actually experience a 72% security incident rate, compared to 33% for those who remain cautious, according to the 2026 Infrastructure Identity Survey.
- Only 13% of organisations feel extremely prepared for the reality of agentic AI despite the majority racing toward autonomous adoption, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: How do organisations decide whether an AI agent should be allowed to act autonomously?
A: Autonomy should be granted only when the business process can tolerate runtime action, the owner is clear, the policy engine can make immediate decisions, and the activity is fully attributable. If any of those are missing, the safer model is constrained delegation with challenge or denial for sensitive steps.
👉 Read our full editorial: Agentic traffic is forcing a new identity security model