What happened
Fastly has expanded its product portfolio with the launch of AI Runtime Control, AI Firewall, and new API security capabilities. These tools are designed to provide real-time visibility and governance for organizations transitioning AI deployments from experimental phases to production environments.
Fastly's new AI Runtime Control sits between applications and AI models, acting as a central routing endpoint for both public and self-hosted providers. It includes features for token spending visibility, rate limiting, and budget controls, while using virtual keys to protect provider credentials. This allows organizations to switch between different model providers while maintaining consistent policy enforcement.
The AI Firewall is specifically designed to protect AI applications by evaluating prompts in the request path before they reach the model, aiming to mitigate risks such as . Additionally, the new API Security features are intended to govern how AI agents interact with enterprise APIs, allowing organizations to enforce API contracts and block non-compliant requests from agentic or agent-assisted traffic.
The company stated that these tools are a response to the rapid growth of machine-generated traffic, which it claims rose above 50% on its network during July and August. Fastly cited McKinsey research indicating that 93% of organizations are currently exceeding their AI budgets, highlighting the need for the cost-management features included in the new release.
Source details: securitybrief.co.nz ↗
Why it matters
As organizations increasingly integrate AI models and autonomous agents into their infrastructure, they face significant challenges regarding cost management, security, and operational reliability. Fastly's new suite addresses these by centralizing policy enforcement and security controls directly in the request path. By providing a unified layer for model routing, token budget management, and mitigation, the company aims to help enterprises scale AI adoption without sacrificing security or exceeding operational budgets.
The shift toward multi-model architectures means enterprises are increasingly reliant on a complex web of external and internal services. Fastly's approach attempts to consolidate security and governance on the same edge platform used for content delivery and DDoS protection, potentially simplifying the security stack for IT teams.
By moving security and control to the edge, Fastly aims to provide the 'real-time' governance that Kelly Shortridge, Chief Product Officer, argues is necessary to maintain business resilience. The ability to observe or block traffic on a service-by-service basis provides a granular level of control that is critical as AI agents begin to automate more complex software development and operational tasks.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').What is the most accurate way to describe what AI Agents can do today?
What to watch next
The effectiveness of these tools in mitigating sophisticated attacks and their ability to handle diverse, multi-vendor AI architectures will be key indicators of their utility. Observers should monitor how well these edge-based controls integrate with existing enterprise workflows and whether they successfully reduce the operational friction associated with managing heterogeneous AI model deployments.
Pricing and specific availability details for these new features were not disclosed in the announcement. Potential users should verify whether these tools are currently generally available or if they are being rolled out in phases.
The long-term impact of these tools will depend on their compatibility with the rapidly evolving landscape of AI frameworks and the specific types of attacks they can effectively neutralize. As AI agents become more autonomous, the ability of these tools to enforce strict API contracts will be a critical test of their efficacy in preventing unauthorized or malformed operations.