AI bots scan TLS certificate logs: any information used to train LLMs

Human Security's State of AI Traffic report reveals a dramatic rise in AI bot activity: AI agent traffic increased 7,851%, AI scraper traffic 597%, and overall AI traffic 187% — growing eight times faster than human traffic. By early 2026, AI traffic officially overtook human traffic, with bots making up nearly 90% of visits to content-rich sites like e-commerce and media. Fake accounts are growing exponentially. Experts call this the end of the open web as content is instantly consumed by bots to train LLMs, causing human traffic to plummet (e.g., tech media visits down over 50%). Publishers are moving content behind paywalls and blocking even Internet Archive. Bots also scan TLS certificate logs (Certificate Transparency logs) to discover new domains for crawling. The report is based on Human Defense Platform data. Cloudflare CEO Matthew Prince predicted this turning point would occur in 2027, but it happened earlier in 2026. Additionally, data brokers now sell archives of internal company communications for LLM training, and firms create 'digital twins' of employees using chat logs, emails, and video recordings to replicate skills if an employee leaves.