TL;DR: 187% in AI-driven traffic and 7,851% in agent-driven traffic show how fast these sessions are growing, from HUMAN's 2026 State of AI Traffic & Cyberthreat Benchmark. Session-level telemetry and trust-based classification are now needed to separate agent behaviour from human and generic bot activity.
Editorial analysis by NHI Mgmt Group, based on content published by HUMAN: “Agentic Visibility: How to See AI Agents in Your Traffic”.
By the numbers:
- AI-driven traffic grew 187% in 2025, according to HUMAN.
Key questions
Q: Why does bot-only traffic analysis fail for AI agents?
A: Because AI agents do not behave like simple crawlers.
Q: What are the signs that visibility is failing in an agentic environment?
A: A common sign is when analysts can describe individual events but cannot reconstruct the full path from initial interaction to downstream impact.
Q: How should security teams govern AI agents that browse and transact on behalf of users?
A: Security teams should govern AI agents as delegated actors with narrow, task-scoped permissions, not as enhanced browsers.
Practitioner guidance
- Define an agent traffic policy model Set explicit handling rules for high-trust, medium-trust and low-trust AI agents, including when they can browse, compare, log in or be blocked.
- Instrument session-level agent telemetry Capture routes, pageviews, search terms and completion signals so each agent session can be reviewed rather than only aggregate traffic counted.
- Separate agent metrics from human analytics Create reporting that isolates AI-driven traffic from human visitors and traditional bots so attribution and journey data remain decision-grade.
Bottom line: AI agents are already large enough in web traffic to distort analytics, attribution and journey analysis when organisations cannot classify them cleanly.
Glossary terms: Agentic Data Visibility, Trust-Based Classification, Session-level telemetry, Non-Human Identity (NHI)
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
Agentic visibility is becoming a governance primitive, not an analytics add-on. When AI agents browse, compare and transact, the organisation needs a distinct control plane for identification, classification and response. Treating agent traffic as just another bot stream hides policy-relevant behaviour and weakens both analytics and fraud governance. Practitioners should define agent visibility as a board-relevant measurement capability, not a convenience metric.
A few things that frame the scale:
- 19% of organisations give AI systems dramatically more access than human employees, nearly one in five granting unrestricted privilege, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: Should teams block all low-trust AI agents by default?
A: Not automatically. Some low-trust traffic is malicious, but some legitimate partner or experimental agents may still need limited access. The better approach is to separate the policy decision from the visibility decision, then apply challenge, throttling or blocking based on the risk of the specific session and flow.
👉 Read our full editorial: Agentic visibility becomes a control gap in AI traffic analytics