GLOSSARY

A technical report outlining factors, threats, and architectural considerations influencing trustworthiness in artificial intelligence systems.

Understanding trustworthiness vulnerabilities helps enterprises anticipate safety incidents, defend against adversarial threats, and build stakeholder trust in automated decisions.

A financial institution conducts a threat assessment on a fraud detection system against ISO/IEC 24028 trustworthiness characteristics, identifying vulnerabilities to evasion attacks and dataset shift.

The standard provides an informational overview of trustworthiness factors and vulnerabilities, but does not specify certifiable requirements or testing methodologies.

The document explores key trustworthiness characteristics, including reliability, availability, resilience, robustness, safety, explainability, transparency, privacy, and fairness. It categorizes vulnerabilities and threats such as data poisoning, adversarial manipulation, model evasion, and bias, offering architectural mitigation approaches.