AI Risk Library
A structured taxonomy of documented AI risks. Each entry covers what the risk is, how it manifests, how to detect it, and what mitigation strategies are available.
Prompt Injection: Definition, Attack Patterns, Detection, and Mitigation
Prompt injection allows attackers to override AI system instructions through crafted inputs. Patterns, detection methods, and mitigations.
Overreliance on AI: Automation Bias, Skill Atrophy, and Organizational Controls
Overreliance on AI systems leads to degraded human judgment and missed errors. Patterns, warning signs, and organizational controls.
AI Bias: Types, Impact Assessment, Detection, and Mitigation
AI systems can encode and amplify existing biases in training data, model design, and deployment context. Assessment frameworks and mitigation approaches.
AI Data Leakage: Patterns, Detection, and Mitigation
Data leakage through AI systems occurs when sensitive information is exposed through model outputs, training data memorization, or API logging. This entry covers five documented patterns, a risk matrix, detection methods, and regulatory implications.
AI Hallucination: Definition, Patterns, Detection, and Mitigation
AI hallucination occurs when a model generates confident, plausible-sounding information that is factually incorrect. This entry covers how it happens, how to detect it, and what mitigation strategies are available.