anomaly
This glossary entry explains anomaly for AI governance and model risk programs. The sections below summarize what the term means in plain language, why chief AI officers and cross-functional committees track it, where teams often get confused, and—when you are signed in—how it shows up across major industries and in expectations tied to the EU AI Act and NIST AI RMF. Use related links at the end of the page to explore neighboring concepts without losing context.
What It Means
An anomaly is when a system behaves differently than expected based on how it's supposed to work according to its design, documentation, or past performance. It's essentially anything that makes you say 'that's not right' when comparing actual behavior to what should be happening. Think of it as a deviation from normal, expected operation that catches your attention.
Why Chief AI Officers Care
Anomalies in AI systems can signal everything from data quality issues to model drift to security breaches, making early detection critical for maintaining system reliability. They often serve as the first warning sign of larger problems that could impact business operations, customer trust, or regulatory compliance. Missing anomalies can lead to AI systems making poor decisions that affect revenue, reputation, or legal standing.
Real-World Example
A credit scoring AI model that typically approves 15% of applications suddenly starts approving 45% of applications over two days, even though the incoming application profiles look similar to historical data. This anomaly could indicate the model has been compromised, the data pipeline is corrupted, or there's been an unauthorized change to the approval thresholds.
Common Confusion
People often confuse anomalies with errors or bugs, but an anomaly is just something unexpected - it might actually be correct behavior in response to changed conditions. The key difference is that anomalies require investigation to determine if they represent problems or simply new normal operating conditions.
Industry-Specific Applications
See how this term applies to healthcare, finance, manufacturing, government, tech, and insurance.
Healthcare: In healthcare, anomalies represent deviations from established clinical protocols, patient safety standards, or regulato...
Finance: In finance, an anomaly refers to unexpected patterns in financial data, trading behavior, or risk metrics that deviate f...
Premium content locked
Includes:
- 6 industry-specific applications
- Relevant regulations by sector
- Real compliance scenarios
- Implementation guidance
Technical Definitions
NISTNational Institute of Standards and Technology
"Anything observed in the documentation or operation of a system that deviates from expectations based on previously verified system, software, or hardware products or reference documents."Source: IEEE_Soft_Vocab
"Condition that deviates from expectations, based on requirements specifications, design documents, user documents, or standards, or from someone's perceptions or experiences."Source: SP800-160
Explore more glossary terms
Discuss This Term with Your AI Assistant
Ask how "anomaly" applies to your specific use case and regulatory context.
Start Free Trial