In the rush to adopt generative AI (GenAI) and large language models (LLMs), many organizations are unaware of just how deeply these technologies are already embedded in their systems. APIs are the silent backbone enabling this integration—but they’re also a growing attack surface.

Rupesh Chokshi, SVP and GM of Application Security at Akamai, joined us to talk about a pressing reality: GenAI is showing up in enterprise environments in ways security teams don’t fully understand.

“Many customers didn’t realize that GenAI was already being utilized inside their org,” Chokshi said. “Until we started to show them that here are dozens of API calls that are taking place.”

Akamai’s response is a new feature called API LLM Discovery. This solution tags and classifies API traffic related to LLMs—identifying the host, path, and function of each call. That level of granular visibility is critical in a world where GenAI services can be accessed with a few lines of code or a plug-in.

But visibility alone isn’t enough. The Akamai platform goes further by integrating this insight with runtime protections, including its Firewall for AI. “The combination of discovery and runtime protection with a focus on APIs is the value proposition we bring forward,” Chokshi explained.

This layered approach ensures that even if a shadow LLM interaction is discovered today, it can be secured tomorrow—with minimal disruption to developers and zero blind spots for security teams.

Why it matters now:
APIs have long been a top target for attackers. With GenAI in the mix, the stakes are even higher. As more enterprises experiment with or deploy LLMs, the risk of data leakage, unauthorized access, or misuse of AI capabilities grows. Yet most security tooling lags behind.

Akamai’s bet? Discovery is step one. Continuous visibility, paired with runtime protection, is how organizations can embrace AI without sacrificing security posture.