Why can ordinary accuracy be misleading for anomaly detection?

A. Anomalies are often extremely rare B. Accuracy cannot be computed for binary labels C. Normal cases are always missing D. Anomaly scores are categorical