In March 2026, the resource TheNumbers.com was subjected to a massive influx of requests caused by a combination of agentic AI activity and the interests of prediction market participants, such as Polymarket. The incident revealed a critical vulnerability of modern web resources to a new type of load — asymmetric crawling.
What Happened
The film data resource TheNumbers.com faced a massive surge in requests that essentially resembled a DDoS attack. The primary cause was asymmetric crawling: while traditional search engines like Google scan an average of about 5 pages per visitor, agentic models (such as Anthropic) are capable of requesting up to 38,000 pages in a single session. This leads to extreme depletion of server resources without providing content owners with corresponding return traffic.
Context
The problem is exacerbated by a shift in traffic patterns: instead of standard search indexing, the main driver of load is active data collection by autonomous AI agents. These agents seek information to execute instantaneous trades on prediction markets, turning ordinary data collection into a resource-intensive process that is effectively free for the consumer, creating an economic imbalance in favor of AI systems.
Why It Matters for the Industry
For the industry, this means a need for a radical overhaul of web resource protection and traffic monetization models. A transition from the advertising model to 'pay-per-scrape' protocols and the implementation of paid API gateways for legitimate scanning (e.g., solutions from Cloudflare) is expected. In the long term, this could lead to the fragmentation of the internet and the formation of 'walled gardens,' where access to high-quality data becomes a strictly regulated market.
Why It Matters for Users
For regular users and content creators, this carries the risk of degrading the availability of useful services. The independent internet is becoming more fragile: the rise in operating costs for hosting and protection against aggressive bots could make maintaining niche and specialized websites economically unviable, ultimately reducing the diversity and stability of the information space.
Sources
Author
Look at AI, Editorial Team