Governing AI Risk

The capstone idea of this track is that defending and deploying AI is not a side project bolted onto security — it is security, extended to a new and fast-moving asset class. The organizations that handle it well do not invent a parallel universe of AI controls; they fold AI risk into the risk and security program they already run, and they use frameworks to do it systematically.

The most widely referenced structure is the NIST AI Risk Management Framework, which organizes the work into functions — broadly, govern, map, measure, and manage — mirroring the lifecycle thinking you saw in the Cyber Defense Matrix track. Govern establishes the policies and accountability. Map identifies where and how AI is used and what could go wrong. Measure assesses those risks. Manage prioritizes and treats them. The value is not that the framework eliminates AI risk — nothing does — but that it gives a cross-functional team a shared language and a repeatable process, the same way NIST CSF does for general cybersecurity.

Risk follows the AI lifecycle, and each stage has its own exposure: the data stage (provenance, poisoning, privacy of training data), the training and fine-tuning stage (integrity, backdoors), the deployment stage (prompt injection, insecure output handling, agent permissions), the operations stage (monitoring for drift, abuse, and the automation-bias failure modes from Module 2), and decommissioning (what happens to the model and its data). A program that secures only deployment while ignoring data provenance has covered one stage of five.

Step back and the AI DR domain reveals itself as the connective tissue of the whole academy. An AI system runs on devices, communicates over networks, lives increasingly in the cloud, is reached through applications, learns from and produces data, and is operated by users with identities. Securing it means applying every other domain's controls and adding the AI-specific defenses from this track on top. That is why AI DR is both the newest domain and the one that ties the others together — and why a professional who understands it can see how an attack chains across the entire defensible city. With this track complete, you hold the modern half of the matrix; the domain tracks fill in the rest.