# AI Security Monitoring and Incident Response (Self-Paced) Canonical URL: ## Overview A tightly scoped, practitioner-level course on the two places conventional IT monitoring and incident response fall short with AI: knowing what a compromise actually looks like, and knowing what to do once you see one. The idea holding it together is that an AI system carries every traditional attack surface and then piles on more, all bound up in data, models and inference behavior — which means an attacker can tamper with what a model learns from, or with what it is shown at run time, and never come near the infrastructure. Monitoring has to move with that, away from "is the system up" and toward "is the system behaving the way it was meant to, on data it was meant to see, for purposes it was approved for." **The practical core is a signals table that maps observed patterns to likely causes** , paired with a first-hour triage sequence. A sustained spike in near-identical unusual inputs reads as adversarial probing; a slow accuracy decline reads as drift; an unexplained query spike from one credential reads as an extraction attempt. From there the course turns to judgment — separating anomalies from incidents, holding on to the evidence responders routinely lose (the model version in use, the prompt and response logs around the event, training data lineage, the guardrail configuration that was in effect), and grading severity by impact rather than technical difficulty. Its sharpest teaching point: a wrong eligibility or pricing decision caused by a data integrity failure is a high-severity incident even though nothing technically broke. The last module runs the four-phase incident response lifecycle with containment built for AI — rolling a model back to a validated version, switching automated decision-making off in favor of human review, quarantining a training data source that has been compromised, and shutting a confirmed prompt injection vector. ## What you'll learn - Explain how AI systems introduce security risks that differ from, or compound, traditional IT risks - Use the working vocabulary of AI security incidents, including data poisoning, adversarial examples, prompt injection, model drift and model extraction - Describe what continuous monitoring means for a high-impact AI system, beyond uptime and throughput - Distinguish the four categories of AI monitoring and select the technique that fits each - Apply a risk-based approach to security event logging for AI systems, and identify what must be captured and retained - Read common monitoring signals and identify what each most likely indicates - Recognize which anomalies qualify as AI security incidents and which do not - Preserve the evidence unique to AI incidents, including model version, prompt and response logs, and training data lineage - Classify incident severity by impact on safety, rights, mission and data rather than by technical complexity - Apply the four-phase incident response lifecycle to an AI-specific scenario - Select containment actions specific to AI systems, including model rollback and disabling automated decisioning - Carry out post-incident activity including root cause analysis and revalidation before returning a system to automated operation ## Prerequisites Foundational cybersecurity knowledge. Familiarity with AI and machine learning terminology is helpful but not required. ## Pricing **Tuition:** $799