How Cisco Defends at Machine Speed with OpenAI

Hi everyone, Shii-chan here! Today I'm covering a Cisco security story published by OpenAI.

OpenAI News openai.com

What was announced?

OpenAI News describes how Cisco uses OpenAI's frontier models to defend against AI-enabled attacks that move at machine speed. It's a customer story, and it includes no specific performance metrics, benchmarks, or dollar figures.

Why it matters

Jeetu Patel, Cisco's President and Chief Product Officer, says that AI hands attackers speed and scale, and that using AI for defense is a prerequisite for staying safe.

If attacks run at machine speed, defense has to keep up. That's the starting point of this story.

What changes

Cisco combines its network intelligence with OpenAI's models:

• Signal correlation: signals that look benign alone can reveal a meaningful pattern together, and the models correlate them at machine scale
• Vulnerability management: the Daybreak and GPT-6 Astra initiatives help find and patch vulnerabilities more efficiently and shorten investigation, response, and remediation
• Fitting existing workflows: designed to work within customers' current scanning workflows without changing how teams work

Dive Deep

The story also covers network-centered defense:

• Security is shifting from a single perimeter to controls spread across microservices, containers, and Kubernetes clusters
• Patel describes putting a small security stack in front of each component and fusing security into the network fabric
• Application-level controls limit each service to the connections it needs, reducing the paths an attacker can use
• Teams can isolate affected systems selectively, so unaffected services keep running during investigation and response

On data handling, OpenAI's enterprise approach rests on zero data retention, strict customer data boundaries, and verifiable isolation.

Cisco and OpenAI are also exploring, through the Daybreak Defense Network, ways to help security teams triage exposure faster and enforce the right controls in Cisco Cloud Control.

For securing agentic AI, Patel names three priorities:

• Protect agents from malicious attacks
• Detect and intercept unintended agent behavior
• Operate in near real time

Teams need to tell malicious interference apart from an agent pursuing a legitimate goal in an unintended way. That takes combining security telemetry with observability data.

Wrap-up

• Cisco strengthens defense by combining network intelligence with OpenAI models
• Signal correlation and more efficient vulnerability fixing are the pillars
• It is designed to fit existing scanning workflows
• Securing agentic AI relies on combining security telemetry with observability data

This is for anyone curious about security operations in the AI era and how to protect agents!