Security Engineer

Sample question 1

Skyharbor Regional Airport's baggage-handling system runs on an industrial control network. A firmware update pushed to the conveyor PLCs last month changed motor-stop timing, but the change was never validated against the emergency-stop interlock logic because the maintenance window was cut short before test scripts could run. Conveyors have operated without incident since. As the security engineer reviewing this change during a post-incident audit, which statement best captures the operational and safety impact?

1. A.The impact is best treated as operational, since the conveyors have run without incident and firmware changes rarely affect the safety interlock logic.
2. B.The interlock may not stop the belt fast enough, so daily operation looks normal while an injury risk stays undetected until a stop is triggered.
3. C.The impact is best treated as a safety concern, since an untested firmware change would not affect throughput or scheduling once belts are running normally.
4. D.There is no meaningful impact until the interlock is actually triggered, so tracking belt uptime is enough to catch any problem before harm occurs.

Sample question 2

At PetroFlow Refinery, a security engineer investigating a suspected intrusion on the distillation unit's OT network confirms an attacker gained access to the VLAN carrying sensor traffic for a critical pump and, in a prior incident, demonstrated the ability to inject fabricated sensor values on that same segment. During the investigation, the plant's Fault Detection and Diagnostics system reports a developing bearing fault on that same pump and recommends an immediate maintenance shutdown. The maintenance team wants to schedule the shutdown right away. What should the security engineer recommend first?

1. A.Approve the shutdown as scheduled, because FDD tools cross correlate multiple independent sensor streams, which makes their fault diagnoses immune to manipulation of any single sensor
2. B.Tell the maintenance team to disregard the fault diagnosis entirely, since FDD output reflects mechanical conditions and has no bearing on an ongoing network intrusion
3. C.Postpone both the shutdown and the incident response until the FDD vendor reviews the diagnostic algorithm's logic and confirms its output is accurate
4. D.Independently verify the pump's physical condition through manual inspection before acting on the FDD diagnosis, since its conclusion depends on sensor inputs that may be manipulated

Sample question 3

During an AI system audit at a healthcare analytics firm, the security engineer finds that the same small team of data scientists can, all under one shared service account credential, pull identifiable patient records into the training data store, fine-tune the clinical-notes summarization model, and approve the updated model for production. Which finding should the engineer prioritize in the risk report?

1. A.The model's summarization outputs are stored without encryption at rest, which could let an attacker who gains network-level access to the storage layer read identifiable clinical notes directly.
2. B.The training data store's password policy does not enforce the firm's standard minimum length of 14 characters, though multi-factor authentication is still required for every login attempt.
3. C.The environment lacks segregation of duties: one shared credential lets the same people create training data, fine-tune the model, and approve its own deployment, with no individual accountability.
4. D.The shared service account has a session timeout of eight hours rather than the 30-minute idle timeout the firm applies to interactive human user accounts on other systems.