Human Security’s State of AI Traffic report, released 26 March 2026 and based on one quadrillion interactions processed through its Human Defense Platform, confirms that machine-generated traffic has surpassed human traffic globally. Automated traffic (defined as software/AI systems rather than human users) expanded almost eight times faster than human activity throughout 2025. AI traffic alone rose 187% from January to December, while agentic AI (autonomous agents performing actions on behalf of users) exploded nearly 8,000% from negligible 2024 levels.
The shift is driven by mainstream adoption of large language models and features such as Google’s AI Overviews and autofill. CEO Stu Solomon stated the foundational assumption of a human on the other side of the screen “is very rapidly being replaced,” with machines now the dominant form of internet traffic. Cloudflare CEO Matthew Prince noted pre-generative-AI bot traffic was ~20% (mostly Google crawlers) and had predicted AI bots would exceed humans only by 2027; the milestone arrived two years early.
Importantly, the surge is largely non-malicious — it reflects legitimate AI usage rather than spam or attacks. However, the report acknowledges measurement limitations: reliance on self-reported user-agent strings is increasingly noisy and self-identification reliability is declining. No single comprehensive database exists, yet the scale of the dataset makes this the clearest benchmark to date of the post-ChatGPT internet.
// Share Your Analysis