Ai overwhelms internet: human traffic now a minority
The internet, once a human-centric space, is rapidly being eclipsed by automated systems. A startling new report from Human Security reveals that artificial intelligence now generates more than 8% of all internet traffic, a figure that has exploded eightfold in just one year. This isn't a gradual shift; it's a wholesale takeover, fundamentally altering the digital landscape as we know it.

The rise of the agent: beyond bots
For years, bots have lurked on the periphery of the web, silently powering advertising and social media. But the current surge isn’t driven by simple scripts. It's fueled by 'agentic ai'—advanced systems like those powering ChatGPT, Gemini, and Claude—capable of independent action and strategic interaction. These agents are not just reacting; they’re actively shaping the flow of information, and they’re doing so at an astonishing rate.
The data is unequivocal: 1.7% of internet traffic is directly attributable to agentic ai, but that figure masks a much larger story. A staggering 65.5% is consumed by ai training crawlers – systems constantly scraping data to refine algorithms – and 31.9% by real-time ai scrapers. Combined, these ai-driven activities account for a staggering 95% of all network activity, concentrated primarily within the retail, media, and travel sectors.
Stu Solomon, CEO of Human Security, notes the unsettling speed of this transformation. “Internet was built on the assumption of a human on the other side of the screen,” he explains. “That assumption is being replaced very rapidly.” OpenAI’s ChatGPT dominates the landscape, responsible for 69% of this agentic traffic, followed by Meta AI (16%) and Anthropic's Claude (11%). The implications are far-reaching, extending beyond mere inconvenience.
The shift signifies a move toward an ‘Internet of AI,’ one where human interaction is increasingly marginalized. Security vulnerabilities and the integrity of information have become critical concerns. The rapid proliferation of these systems isn't about a dystopian showdown between humans and machines; it’s about establishing boundaries. We need to ask: who is setting the rules for this new digital order, and are those rules aligned with human interests?
Forget about reliable Wi-Fi. The future, it appears, is wired – literally. Human Security's findings suggest a move towards mandatory Ethernet connections, a desperate attempt to maintain some semblance of control in a network rapidly slipping from our grasp. The question isn’t whether AI is changing the internet, but whether we can adapt before it remakes it entirely in its own image.
