· ai agents

AI agents are the majority of your traffic now. Can you prove it?

AI agents crossed a line in 2026. On June 3, 2026, Cloudflare's CEO Matthew Prince shared Radar data showing automated requests had overtaken humans for the first time in internet history: bots generated 57.5% of HTML web traffic, against 42.5% from people. AI crawlers are the fastest-growing part of that majority, and most sites cannot name a single one of theirs.

AgentBlock is Ingenious AI's classification and blocking kit for known AI agents: a Cloudflare Worker, WordPress plugin or JS snippet that acts on your policy at the edge. This check starts one step earlier: could you even identify the agent traffic you already get?

Honest scope, read this first: this page reports Cloudflare's published network measurements as reported; your site's mix will differ, which is the point of checking. AgentBlock classifies traffic by known agent user-agents and enforces your allow or block policy. A scraper that forges a browser user-agent defeats classification, and no honest tool claims otherwise.

AI agents and the majority flip of June 2026

Cloudflare Radar sits on enough of the web to call the turn: on June 3, 2026, its CEO announced bots had passed humans, 57.5% of HTML traffic to 42.5%. Imperva's 2025 Bad Bot Report had already put automated traffic at 51% of all web traffic for 2024. The direction is not in dispute, and the composition matters: AI crawlers reached roughly a fifth of verified bot traffic in 2026 measurements, with training crawlers the largest slice.

Sites experience this as a quiet distortion. Analytics show the human minority and call it 'traffic'. Server bills, crawl budgets and content exposure are set by the machine majority nobody itemized. The first governance act is simply naming your agent visitors: which companies, which crawlers, what volume, allowed or not.

57.5%of HTML web traffic was bots in June 2026 (Cloudflare Radar)
51%automated share of all web traffic already in 2024 (Imperva Bad Bot Report)
5.89%of websites block GPTBot, the most refused AI crawler (Ahrefs)
42.5%the human share: the minority your analytics calls 'traffic'

Why can't you see most of your own traffic?

Because the measurement stack was built for people. Analytics tools run a script in a browser; bots do not run scripts, so the entire machine majority is filtered out before reporting begins. What remains is presented as 'your traffic' without the caveat that it is the human 42.5%. Server logs and edge data are the only complete record.

Check your agent-traffic visibility

Answer for your main site, honestly, as of today. Everything runs in your browser; nothing is sent anywhere.

Take back the majority view: the working order

  1. Itemize the agents: one month of server logs, grouped by AI crawler, with volumes. This list is the foundation of every later choice.
  2. Classify by intent: training, AI search, and transactional agents are different guests. Give each class its own answer.
  3. Publish, then enforce: robots.txt states the policy for honest agents; edge rules enforce it for the rest. AgentBlock packages the enforcement.
  4. Monitor the mix: a monthly look at agent share and growth. What doubles quietly becomes next year's cost line.
  5. Revisit quarterly: new agents launch and old ones change behavior. A policy is a subscription to a changing web, not a one-time setting.

AgentBlock, by Ingenious AI, classifies known AI agents in your traffic and blocks the ones your policy denies, via a Cloudflare Worker, WordPress plugin or JS snippet, with no user accounts by design: your policy, your edge, your call. See how AgentBlock works.

FROM INGENIOUS AI
AgentBlock names the machine majority in your traffic, and lets you decide, agent by agent, who gets served.
All 21 platforms: DisclosureGuard  ·  ModelMinder  ·  TermsWatch  ·  Vibe-Code Audit  ·  Rogue-Agent Defense  ·  BreachHunter  ·  ChatBreaker  ·  Agent Flight Recorder  ·  Agent Auth Gateway  ·  DomainGuard  ·  ContentCraft AI  ·  EvalScope  ·  BoundaryVault  ·  Agentic Receipts  ·  Agent-Ready Store Audit  ·  AgentBlock  ·  HomeShield MCP  ·  AgentGate  ·  ADMT  ·  DueLadder  ·  WireWatch
One suite, one deal: the Ingenious AI Security Suite. All 21 platforms are live now:
Chat with us on WhatsApp →

Quiz: how well do you know the traffic flip?

1. What did Cloudflare report on June 3, 2026?

2. What did Imperva measure for 2024?

3. Why do analytics dashboards miss bot traffic?

4. Which AI bot is the most widely blocked?

5. Where does a complete record of your agent traffic live?

Frequently asked questions

What are AI agents in web traffic?

In web traffic, 'AI agents' covers the automated visitors from AI systems: training crawlers collecting content, AI search crawlers fetching pages to answer questions, and transactional agents browsing or buying for users. Together with other bots they passed human traffic in 2026: 57.5% of HTML requests, per Cloudflare. AgentBlock classifies the known AI ones by user-agent.

What is AgentBlock?

AgentBlock is Ingenious AI's one-click kit for governing known AI agents on your site. Pick a policy, deploy it as a Cloudflare Worker, WordPress plugin or JS snippet, and denied agents are blocked at the edge. It deliberately has no user accounts: the classification and policy run in your infrastructure, under your control.

How much of my traffic is AI agents?

It varies, and the honest answer is: measure it. Network-wide, bots were 57.5% of HTML traffic in June 2026 (Cloudflare), and on small-business sites AI crawlers alone made 40.7% of identified bot requests (LovedByAI, 924 sites). Your mix is in your server logs, grouped by user-agent, waiting to be counted.

Why not block every bot?

Because bots are not one thing. Search crawlers send your customers, AI search crawlers increasingly do the same, and transactional agents may arrive ready to buy. Training crawlers are the usual block target: they take content and return nothing. A split policy beats a blanket one in both directions.

Can AgentBlock stop a scraper pretending to be a browser?

No, and be suspicious of any tool that promises it. Classification works on declared identity: known agent user-agents. A scraper that forges a browser identity needs other defenses, such as rate limits and behavior analysis. AgentBlock's honest scope is the declared majority, which is most of the volume and all of the reputable crawlers.

Sources: Cloudflare Radar, automated traffic milestone (June 3, 2026, via Cloudflare CEO Matthew Prince; compiled at digitalapplied.com) · Imperva 2025 Bad Bot Report · Ahrefs AI-bot blocking data (via superlines.io) · Ahrefs keyword data (ai agents, 46,000/mo US).

Related checks