Automated traffic has officially surpassed human activity online. A new report from cybersecurity firm Human Security, covered by Hacker News, confirms what many suspected: AI and bots now generate more internet traffic than people do.
The numbers are striking. According to Human Security’s State of AI Traffic report, automated traffic grew almost eight times faster than human activity in 2025. AI-specific traffic jumped 187% between January and December of that year alone. And agentic AI traffic, where autonomous agents perform tasks on behalf of users, exploded by nearly 8,000%.
“The internet as a whole was created with this very basic notion that there’s a human being on the other side of the computer screen, and that notion is very rapidly being replaced,” said Stu Solomon, CEO of Human Security.
What’s driving the surge
Three forces are pushing this shift:
- LLM proliferation: ChatGPT, Claude, Gemini, and their competitors are generating massive volumes of queries, crawling, and data retrieval
- Agentic AI: Tools like OpenClaw that act autonomously on behalf of users went from near-zero volume in 2024 to significant traffic in 2025
- Embedded AI features: Google’s AI Overview, autofill, and similar baked-in automation now account for a growing share of web requests
This isn’t just about chatbots answering questions. It’s a fundamental shift in who (or what) is consuming the web.
A caveat worth noting
The report has limitations, and it’s honest about them. Human Security’s data comes from its own Defense Platform product, which processed over one quadrillion interactions across its customer base. That’s a lot, but it’s not the whole internet.
Filippo Menczer, a professor of Informatics and Computer Science at Indiana University, pointed out the measurement challenge: “You can try to estimate the amount of bot traffic by looking at the agent strings, but these are very noisy estimates. They depend on what sample you get.”
Human Security itself acknowledges that user-agent string self-identification, the primary method for spotting AI crawlers, is becoming less reliable.
Why this matters
Solomon made an important distinction: automated traffic isn’t inherently malicious. “This notion of machine bad, human good just is not realistic,” he said. “You have to live in a world where machines are acting on our behalf.”
That’s the real story here. The internet is transitioning from a human-to-human network to a machine-mediated one. AI agents browsing, summarizing, purchasing, and negotiating on our behalf will only accelerate this.
Cloudflare CEO Matthew Prince offered a timeline at SXSW last week. He said pre-AI bot traffic sat around 20%, mostly from Google’s web crawler. His prediction: AI bots will exceed human traffic by 2027, driven by “the rise of generative AI and its just insatiable need for data.”
Given Human Security’s findings, that 2027 target might already be in the rearview mirror for some corners of the web.
What comes next
Infrastructure, authentication, and trust models built for human users will need rethinking. Website owners face growing pressure to decide which bots get access and which don’t. Content creators are already grappling with AI scrapers consuming their work without compensation.
The internet didn’t stop working when bots took the majority. But the rules of engagement are changing fast, and most of the web’s infrastructure hasn’t caught up yet.
Full details on the report are available at the original source.