Fastly Launched AI Security and Control Tools
The platform introduces runtime governance and firewall capabilities to manage growing automated web traffic.
Updated on Sept. 21, 2026 in Artificial Intelligence
Live Poll
Do you trust that businesses have sufficient control over the AI models and agents they deploy?
Fastly has launched AI Runtime Control, AI Firewall, and API Security capabilities to help organizations manage model access and costs in production. The suite evaluates prompts in the request path to apply security and routing policies across different model providers.
Why it matters
As machine-generated traffic now exceeds 50% of total network volume, organizations require centralized tools to govern model access and address the 93% of companies currently exceeding their AI budgets. These features enable teams to enforce security and spending policies at the edge during rapid production scaling.
The system monitors traffic by routing model calls through a single endpoint, allowing for the application of security and access policies directly in the request path. This contrasts with traditional perimeter-based security by moving governance to the edge of the network.
The players
Fastly
An edge cloud platform provider that offers programmable delivery, security, and computing services for global applications.
The details
The platform functions by intercepting AI-bound requests before they reach model providers. By placing security evaluation directly in the request path, the system validates prompts and applies routing logic dynamically. This architectural change allows administrators to enforce budget and safety constraints without modifying underlying application code.
Timeline
January 2026 through May 2026: AI traffic grew 6.5 times faster than human traffic.
July 2026 through August 2026: Machine-generated traffic crossed 50% of Fastly network traffic.
September 21, 2026: Fastly announced new AI security and control capabilities.
The Tech Race
This development marks a move by edge computing providers to integrate AI-specific governance into their core infrastructure stacks. It follows the accelerating trend of machine-generated traffic overtaking human interaction on global content delivery networks.
Organizations can begin integrating these controls into their production stacks immediately to monitor model usage and enforce spend limits. Engineering teams will primarily use these tools to replace fragmented, manual monitoring workflows with unified endpoint routing.
The takeaway
The rapid expansion of AI-driven traffic necessitates a shift from manual oversight to automated, edge-based policy enforcement. Watch how these security benchmarks compare to industry-standard model performance metrics in upcoming quarterly infrastructure reports.
Further reading
Explore more analysis on this topic in our Artificial Intelligence section.
Live Poll
Do you trust that businesses have sufficient control over the AI models and agents they deploy?
