The agentic internet exposes a governance gap in web traffic

At a glance

What this is: Netacea argues that the rise of agentic web traffic has created a governance gap, with machine actors now operating at a scale and speed that legacy bot controls and analytics cannot classify cleanly.

Why it matters: IAM, PAM, NHI, fraud, and security teams need a shared model for machine traffic because classification, policy, and accountability now affect revenue, abuse prevention, and access governance at the same time.

By the numbers:

• AI bot visits to human visits shifted from 1:200 to 1:31 in a single year.

Context

The governance gap here is not simply more automation. It is that web properties now receive a mix of human users, benign machine activity, LLM scraping, agentic browsers, and third-party agents, while most controls were built to treat traffic as either human or malicious automation.

In practice, that means classification matters before enforcement. If teams cannot distinguish declared machine activity from undeclared scraping or transaction automation, they cannot assign the right policy, risk owner, or commercial response. The article’s starting position is typical for modern digital estates: visibility lags the traffic model.

Key questions

Q: What breaks when web traffic policies were written only for humans and simple bots?

A: Policies built for a human or basic-bot model break when traffic includes LLM scrapers, agentic browsers, and partner automation that looks legitimate at first glance. The failure is classification, because teams cannot assign the right control, owner, or response path without knowing who or what the actor is. That leaves abuse, revenue leakage, and false positives unresolved.

Q: Why does declared versus undeclared machine traffic matter for governance?

A: Declared traffic can be governed through identity, contract, and policy, while undeclared traffic must be treated as an uncertainty and monitored more aggressively. The distinction changes whether a request is acceptable automation, a commercial relationship, or an abuse case. Without it, teams cannot make consistent decisions about access, challenge, throttling, or escalation.

Q: How should organisations measure whether agentic traffic controls are working?

A: Measure whether the platform can classify traffic accurately enough to link each actor type to a policy outcome and a business impact. Good control shows up as fewer false blocks, clearer attribution, better abuse detection, and reduced commercial leakage. If the team still discusses bots only in aggregate, governance is not working.

Q: Who should own governance of non-human web traffic?

A: Ownership should sit across security, fraud, and digital commerce, because the risk spans access governance, abuse prevention, and revenue protection. Security can define classification and control requirements, but product and commerce teams must help set acceptable automation thresholds. If one function owns it alone, the policy will be too narrow for the real risk.

Technical breakdown

Agentic traffic composition and why legacy bot controls fail

The Agentic Traffic Composition Model is a classification lens for machine actors on the web. It matters because not all non-human traffic is the same. Some agents are declared and negotiated, some are scraping content, and some are executing tasks with browser-like behaviour that looks human at the session layer. Traditional bot management often keys off signatures, rate limits, and simple heuristics. Those controls struggle when the traffic is adaptive, distributed, and economically motivated rather than obviously malicious. The result is a governance blind spot, not just a detection gap.

Practical implication: classify traffic by actor intent and business effect before applying blocking or challenge controls.

Declared versus undeclared machine traffic

Declared traffic is machine activity that a platform can identify, attribute, and govern. Undeclared traffic is any non-human activity that arrives without clear disclosure or lifecycle controls. That distinction matters because the same request pattern can carry very different risk depending on whether the actor is a known partner integration, a sanctioned agent, or an unidentified scraper. Once traffic is treated as an identity and policy problem, the control surface expands from network filtering to authorisation, contractual terms, and monitoring. In identity terms, this is a non-human governance problem, not a pure web security issue.

Practical implication: maintain separate policy paths for identified agents, third-party automation, and unknown machine traffic.

Why economic impact is the right control objective

The article frames governance around economic impact rather than guesswork because that is where these actors create harm. Scrapers can distort analytics, consume infrastructure, degrade user experience, and extract value from content or commerce flows. Agentic browsers can also move from passive retrieval into actions that affect inventory, bookings, and transactions. In that context, a purely technical success measure such as block rate is incomplete. The better question is whether the platform can protect margin, trust, and service integrity while allowing acceptable automation.

Practical implication: tie traffic policy to revenue, fraud, and service integrity outcomes instead of treating all automation as one category.

NHI Mgmt Group analysis

Agentic web traffic is becoming an identity governance problem, not just a bot-management problem. Once machines can browse, scrape, and transact, the question shifts from whether traffic is automated to whether it is governed as a distinct class of non-human activity. That creates an identity boundary problem because policy must follow actor type, intent, and lifecycle. Practitioners need to treat machine traffic as part of the access governance model.

Visibility has become the prerequisite control for agentic commerce. The article’s central insight is that declared and undeclared machine traffic create different risk profiles, so enforcement before classification produces the wrong outcomes. A platform cannot set fair policy, measure abuse, or assign accountability if it cannot tell a legitimate integration from an opportunistic agent. The practical conclusion is that governance starts with classification quality.

Agentic traffic composition: the new control concept is to classify non-human web actors by intent, disclosure, and business effect. That framing is more useful than a binary human-versus-bot model because it reflects how modern traffic actually behaves. It also creates a common language for security, fraud, commerce, and platform teams. Practitioners should build policy and telemetry around this composition model rather than around legacy bot labels.

Security teams cannot govern this problem alone because the consequences are commercial as well as technical. The article correctly places revenue capture, fraud, competitive protection, and compliance in the same frame. That means ownership will have to span security, digital commerce, and risk functions. The practitioner lesson is to define shared policy authority before machine traffic volume forces the decision by default.

From our research library:

• AI-related credential leaks surged 81.5% year-over-year in 2025, with the surrounding AI infrastructure leaking 5x faster than core LLM providers, according to the State of Secrets Sprawl 2026.
• Read next: Agentic AI Identity Guide

What this signals

Agentic traffic composition is now a programme design issue, not a tooling tweak. The immediate change for practitioners is that bot management must move from blocking unknowns to governing classes of machine actors. That means policy, telemetry, and commercial rules need to line up before volume forces reactive decisions.

Machine traffic governance will increasingly sit at the intersection of IAM and digital fraud control. As non-human actors begin to transact, identity assurance for integrations, agents, and partner automation becomes part of platform risk management. Teams that already manage NHI, federated access, and third-party lifecycle will be better placed to absorb the shift.

As machine traffic becomes more economically significant, the platform owner will need a clearer rule for when automation is a customer, a partner, or an abuse case. That rule should be explicit enough to support enforcement, billing, and escalation without ambiguity.

For practitioners

• Define a machine-traffic classification model Separate human users, declared automation, third-party agents, and unknown scrapers into different governance classes before policy enforcement.
• Instrument visibility before blocking Build telemetry that captures actor identity, session pattern, request intent, and downstream business effect so teams can distinguish abuse from legitimate automation.
• Assign joint ownership across security and commerce Set explicit decision rights for bot policy, pricing impact, fraud response, and partner automation so no team is left governing agentic traffic alone.
• Use economic impact as the policy threshold Tune controls to protect revenue, service quality, and content value rather than relying only on request rate or user-agent signatures.

Key takeaways

• The core problem is no longer just bot volume. It is that web platforms must distinguish human users, declared automation, and undeclared agents before policy can work.
• Netacea’s research shows AI bot visits rising from 1:200 to 1:31 in a year, which makes visibility a governance prerequisite rather than a reporting metric.
• Teams need shared ownership across security, fraud, and commerce so that machine traffic policy protects revenue, service integrity, and legitimate automation at the same time.

Key terms

• Agentic Traffic: Traffic generated by software that can act on behalf of a user or process with some degree of independent decision-making. In fraud prevention, it includes both legitimate assistants and malicious automation, so the control question becomes intent and behaviour, not automation alone.
• Declared Traffic: Non-human traffic that identifies itself through known credentials, registration, headers, or contractual context. Declared traffic can still be risky, but it is easier to permit, meter, and audit because the organisation has a stated basis for trust and control.
• Undeclared Traffic: Machine activity that does not clearly reveal who it is, why it is present, or what it is allowed to do. It often mimics normal sessions or rotates identities, which makes enforcement, attribution, and abuse response materially harder.
• Agentic Traffic Composition Model: A governance framework for grouping non-human web traffic by actor type, trust level, and commercial effect. It helps teams decide whether traffic should be allowed, constrained, challenged, or denied based on measurable business and security outcomes.

Deepen your knowledge

NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, agentic AI identity, and machine identity security for practitioners who need to govern non-human access. It helps identity and security teams turn emerging machine actors into a managed programme rather than a blind spot.