Thinking Outside the Black Box: Defenders Need Open Source AI

In the new age of AI-driven attacks, accelerating cybersecurity capabilities to machine speed requires open source models. Open source software speeds response times, expands the global cybersecurity workforce, and provides teams with proven defensive tools.

The recent Hugging Face security incident underscores that need. When closed AI tools tripped up the defenders who were trying to stop the active intrusion, they had to turn to an open-weight model to contain the threat.

Open models and open harnesses benefit from the wisdom of the crowd by democratizing advanced defensive capabilities, enhancing transparency, and allowing organizations to use AI without surrendering control of their most sensitive data. The strength of open source is a massively distributed defense framework managed by a global community rather than by a single technology vendor.

An open, secure, and distributed internet is in Akamai’s DNA. We don't just participate in the ecosystem; we defend it.

Akamai sits at the edge of the global internet, analyzing trillions of daily requests and blocking billions of cyberthreats every single day. From market-leading web application firewalls (WAFs) and API security to Zero Trust microsegmentation — including hardware-isolated protection powered by NVIDIA DPUs — we protect the world's top brands where attacks hit first: at the edge.

But defense is only half the job. We also build the foundation.

In 2025, we signed a multi-year agreement to host and run the Linux kernel's distribution infrastructure, providing Git environments and developer access at no cost to the Linux Kernel Organization, alongside dedicated infrastructure for Alpine Linux. As our Chief Architect put it: “Akamai depends on Linux just like the rest of the world, and this is us giving back.”

That open philosophy runs all the way up our stack. Akamai Cloud is Kubernetes native from the ground up. Our AI serving layer relies on open runtimes that the community can actively inspect and shape — from virtual large language models (vLLMs) and KServe to the NVIDIA runtimes our customers trust.

Our edge native serverless platform draws directly from open Wasm projects in the cloud native ecosystem, while the identity and workload-attestation primitives that the agent stack depends on — like SPIFFE and SPIRE — are the exact open foundations that we build against every day.

For us, open source was never a procurement checkbox: It is the very fabric Akamai Cloud is made of.