Which term describes the set of indicators used to monitor potential AI risks and detect emerging threats?

Which term describes the set of indicators used to monitor potential AI risks and detect emerging threats?

• A. KRIs in AI risk management
• B. Red teaming in AI systems
• C. Data quality in AI security
• D. Explainability in AI

Answer: (A)

Key Risk Indicators (KRIs) in AI risk management are the metrics used to monitor risk exposure and detect early signs of trouble. They track ongoing signals that risk levels may be rising, such as data drift or distribution shifts, declines in model performance, data quality issues, unusual data or access patterns, and governance or policy violations. By watching these indicators, teams can intervene promptly—retraining models, remediating data, adjusting controls, or escalating actions before problems escalate. Red teaming is about actively probing systems to reveal vulnerabilities through simulated attacks, not a standing set of ongoing risk signals. Data quality focuses on the integrity of the inputs themselves, which is important but doesn’t constitute the continuous risk-monitoring framework. Explainability addresses how and why a model makes decisions, rather than monitoring risk indicators.

Key Risk Indicators (KRIs) in AI risk management are the metrics used to monitor risk exposure and detect early signs of trouble. They track ongoing signals that risk levels may be rising, such as data drift or distribution shifts, declines in model performance, data quality issues, unusual data or access patterns, and governance or policy violations. By watching these indicators, teams can intervene promptly—retraining models, remediating data, adjusting controls, or escalating actions before problems escalate. Red teaming is about actively probing systems to reveal vulnerabilities through simulated attacks, not a standing set of ongoing risk signals. Data quality focuses on the integrity of the inputs themselves, which is important but doesn’t constitute the continuous risk-monitoring framework. Explainability addresses how and why a model makes decisions, rather than monitoring risk indicators.