AI Agents Reshape the Internet: Traffic Surge, Security Risks, and Emerging Governance

An Imperial College, Stanford and Internet Archive analysis found over one-third of new sites are AI-generated. Cloudflare estimates a third of traffic is bot-driven, while bunny.net’s 2026 Edge Security Report recorded automated requests at 20 % of traffic during a 53 % rise. AI crawlers accounted for 0.54 % of requests, edging out search bots (0.50 %), and ByteDance’s Bytespider contributed half of AI-crawler traffic.

There Are Officially More AI Agents Than Humans On The Internet, Reports FindLe FigaroAgentOps: The Dawn Of The Internet Of Agents

bunny.net logged a DDoS peak of 16.5 million requests per second from 392 000 IPs and a 4.2 Tbps volumetric attack. Injection-type threats—SQL injection, XSS, remote code execution—made up 45.5 % of firewall blocks; Log4Shell triggered over 184 000 events. Analysts note that rising automated traffic lowers attacker barriers and magnifies breach scale.

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

Enterprises are adopting AgentOps, a framework that adds monitoring and audit to AI agents to curb rogue behavior. The French startup Moderering uses AI agents to block illegal content—hate speech and child sexual-abuse material—in line with the EU Digital Services Act. Surveys show 12 % of SMEs use AI, underscoring a need for data quality, governance and human oversight.

Le Figaro

Dejan Grofelnik Pelzel, bunny.net’s CEO, called automated traffic the new foundation of internet operation. Joe Connolly, bunny.net’s head of Content Delivery & Security, warned that server-based security models are obsolete and urged protection. Alexandre Sossou, Moderering’s founder, highlighted the system’s ability to automatically detect illicit videos and the necessity of human supervision.

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

Security researchers say the automated footprint lowers entry barriers for attackers, making DDoS and injection attacks easier, while critics of AgentOps warn that without oversight agents could act unpredictably. SME advocates caution that AI-driven errors can disproportionately harm small firms, reinforcing the need for human review.

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

Studies use metrics: AI-generated sites exceed 33 % while automated requests are reported at 20 %; AI-crawler traffic is cited as 0.54 % of total requests and as half of bot traffic (Bytespider). No single definition of “automated” traffic is applied, limiting comparison.

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

• “Automated traffic is no longer the exception, it’s becoming the foundation of how the internet operates,” — Dejan Grofelnik Pelzel, CEO, bunny.net

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

• “ Joe Connolly, Head of Content Delivery & Security, added, “The findings from this report clearly show that isolated, per-server security models are no longer fit for purpose against modern traffic patterns.” — Joe Connolly, Head of Security, bunny.net

There Are Officially More AI Agents Than Humans On The Internet, Reports Find

• “We discovered that we had automatically identified websites hosting rape videos,” — Alexandre Sossou, Founder, Moderering

Le Figaro

• “Turning AI agents loose without a plan in place to audit their behaviour is akin to handing teenagers credit cards and not looking at the resulting account statements.” — Analyst, AgentOps

AgentOps: The Dawn Of The Internet Of Agents

Edge security firms plan to expand mitigation for AI-driven attacks, while AgentOps platforms will broaden to manage agent fleets. European regulators are testing AI moderation tools for wider DSA enforcement, and SMEs are expected to raise AI adoption as governance matures.

AgentOps: The Dawn Of The Internet Of Agents