Akamai Launches Secure AI Agent Sandbox and Inference Beta

On October 27, 2026, Akamai Technologies announced the launch of a new AI agent sandbox designed to enhance security for autonomous systems. By utilizing WebAssembly (Wasm) technology, the platform allows developers to isolate agentic code, preventing unauthorized access or malicious execution within enterprise environments. Alongside the sandbox, Akamai introduced a beta program for a 12-model inference service, providing developers with diverse foundation models to power their applications. This development is significant as it addresses the growing need for secure, sandboxed environments where AI agents can operate without compromising the integrity of the host infrastructure. By providing a controlled execution space, Akamai aims to accelerate the deployment of agentic AI in production settings, ensuring that autonomous tasks remain within defined security boundaries. The move reflects a broader industry trend toward 'AI-native' security, where protection is integrated directly into the execution layer of the AI stack rather than being treated as an afterthought.