The internet is now dominated by bots.
Artificial intelligence and bots have officially surpassed human users, according to Human Security’s State of AI Traffic study, which was released on Thursday. The fundamental idea that there is a human being on the other side of a computer screen was used to develop the internet as a whole, but it is quickly being replaced, according to Stu Solomon, CEO of Human Security.
Human Security defines automated traffic as “internet traffic generated by software systems (including AI) rather than human users,” and it has increased dramatically as more people utilize AI chatbots for everyday inquiries.
The paper states that in 2025, automated internet traffic increased nearly eight times faster than human activity.
According to the research, AI traffic rose 187% between January and December 2025, largely due to the proliferation of large language models like Google’s, Anthropic’s Claude, and OpenAI’s ChatGPT. On the other side of the internet, machine-based traffic is essentially taking the place of human traffic, according to Solomon.
The information included in Human’s research came from its Human Defense Platform product, which it claims handled more than one quadrillion client encounters. However, since there isn’t a single comprehensive database of interactions, it can be difficult to assess automated activity over the whole internet.
According to Filippo Menczer, an Indiana University professor of informatics and computer science, you can attempt to estimate the volume of bot traffic by looking at the agent strings, but these are very noisy estimates. They rely on the sample you receive. They rely on the source of the measurements and data.
Although AI operators were identified using user-agent strings, Human’s report notes that “the reliability of that self-identification is a growing concern.”
The increase in agentic activity from AI agents like OpenClaw that carry out tasks on their own for a user is also covered in the paper. Human Security had an almost 8,000% increase in traffic in 2025, despite the incredibly low agentic volume in 2024.
Popular tools like Google’s AI Overview and autofill are examples of automated traffic, which is not always harmful.
As per Solomon, the idea that humans are good and machines are bad is simply unrealistic. We must build a degree of trust that endures over time in a future where machines operate on our behalf.
Despite its shortcomings, the Human Security study is an important standard in the internet’s AI era.
Since ChatGPT’s 2022 launch, the industry has monitored the consistent increase in automated traffic.
Last week at the SXSW conference in Austin, Cloudflare’s CEO stated that previous to the generative AI era, the internet had around 20% bot traffic, powered primarily by Google’s web crawler.
By 2027, AI bots would surpass human traffic, according to Prince, who cited “the rise of generative AI and its just insatiable need for data.”
