MUNICH. The EDGE cloud platform Fastly has published its Threat Insights Report for the second quarter of 2025. The analysis shows that AI crawler-i.e. automated programs that search the Internet for the training of AI systems-make up 80 percent of the entire AI bot traffic. More than half of it goes back to Meta, while Google and Openaai are clearly behind it.
In addition to crawlers, fetcher bots, including those of chatt and perplexity, also contribute significantly to real-time traffic. Over 39,000 inquiries per minute were measured at peak times. This burden takes unprotected servers and networks and has similar effects to DDOS attacks, even without malicious intent.
The report also emphasizes that North America, with almost 90 percent, records the largest proportion of AI Crawler traffic. Other regions such as Europe, Asia and Latin America remain significantly, which, according to Fastly, could indicate a geographical distortion for training data.
Arun Kumar, Senior Security Research at Fastly, explained: “AI-bots are fundamentally changing how the Internet is used and experienced. And bring new complexities for digital platforms. Whether when scraping for training data or in the provision of real-time answers-these bots represent new challenges for visibility, control and costs. Verification standards threatens to become AI-controlled automation to become a blind spot for developer and security teams. ”
In its report, Fastly calls for more transparency in bot verification, a clear labeling by operators and more targeted strategies in bot management. Without these measures, there were increasing risks due to hidden automation, gaps in attribution and increasing infrastructure costs. (Red)
The full threat in Insight Report for the second quarter of 2025:
