- DataDome Analysis Claims Agent Traffic Increased 45% in Q2 2026
- AI metabots have grown more than 163% compared to the previous quarter
- The analysis was performed by the DataDome agent control and bot management platform.
If you run a website, each crawl costs bandwidth, resources, registration, and creates CDN transactions, and while search engine crawlers offered the promise of sending visitors, AI bots do not.
Analysis of a report by cybersecurity company DataDome has shown that while Meta AI bots have increased their activity, they are not generating any significant return to websites.
In contrast, ChatGPT trackers have reduced traffic, but are sending more referrals.
AI agent traffic is growing
While Meta AI is generally considered to be the “chatbot within Facebook,” it appears to be becoming something more, and the emergence of the Meta-WebIndexer bot (which grew 163% in the first quarter) suggests that Meta may be indexing a library of websites, much like Google has done for the past few decades.
The growth of Meta AI as an active tracker is only part of the story, as is the comparative efficiency of ChatGPT. The OpenAI tool seems to know enough about websites that it can provide the answers it already “knows.” On the contrary, the activity of Meta AI that indexes the web seems to explain its strong impact in the second quarter of 2026.
But the Model Context Protocol (MCP) signal is also emerging, which connects AI agents with external tools and differs from standard tracker traffic.
“The second quarter showed us that the terrain is changing faster than most organizations realize. Meta now dominates AI traffic on our network, MCP traffic has emerged as a real signal, and ChatGPT is generating more referral value with fewer traces,” said Jérôme Segura, vice president of threat research at DataDome.
Differences in the way AI agents interact with websites (some behave like users, others pull content) mean organizations must act accordingly.
“What the data makes clear is that not all agents are created equal. Organizations that build policies around these distinctions are the ones that gain an advantage, and that’s exactly why the adoption of trust in agents is accelerating.”
Unfortunately, the existence and growth of MCP into a significant and measurable quantity means that it must also be treated as part of an organization’s attack surface.
Assign resources
Given the origins of the report, there is naturally a cybersecurity aspect to this. While ransomware, malware, and phishing are not going away, autonomous software on the web needs to be approached in a different way.
Those “policy-making organizations” that Segura mentions could, for example, grant full crawl access to Google, allow ChatGPT to retrieve results, but limit the speed of Meta AI based on its poor performance.
Meanwhile, unknown agents (perhaps cybersecurity threats) would require additional verification or be blocked entirely.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
