By sharon
Preferred label Alternative label Hidden label --------- af ar be bg ca cn cs cy da de el en es et fa fi fr ga gl hi hr ht hu id is it iw ja ko la lt lv mk ms mt nl no pl pt ro ru sk sl sq sr sv sw th tl tr uk vi yi × delete
--------- A biased hiring tool caused discrimination due to poorly defined requirements. Access control policies Adaptive retraining schedules Adversarial Attacks Adversarial Attacks Adversarial training A financial AI model exposed client data during exploratory analysis. AI-driven decisions led to regulatory violations in data privacy. AI Life cycle Stages Algorithmic Bias An AI image classifier was fooled by adversarial inputs, causing security lapses in surveillance. Anomaly detection systems API access was abused to extract and replicate a proprietary recommendation model. A retail forecasting AI failed due to unmonitored data drift, leading to supply chain disruptions. Backdoor Attacks Biased decision-making Biased outcomes Bias checks Bias Introduction Clear requirement definitions Compliance failures Compliance frameworks Compliance Issues Compromised model integrity Compromised model integrity Compromised model performance Concept Drift Concept drift detectors Consumer Distrust Continuous monitoring Continuous performance monitoring Data anonymization Data breaches Data Collection and Preparation Data Drift Exploitation Data Exploration and Analysis Data leaks Data leaks Data Leaks Data Poisoning Data Recycling Data regularization Data validation between iterations Data validation processes Degraded model accuracy Discriminatory outcomes Documented cases of Autopilot failures Dynamic compliance audits Encryption Ethical impact assessments Ethical Misuse Ethics, Compliance and Governance Exposed endpoints Governance boards Hyper-parameter Exploits Inadequate data encryption Inadequate monitoring of model performance Inadequate retraining protocols Incidents Incidents Incidents Incidents Incidents Incidents Incidents Incidents Incidents Incidents Incomplete stakeholder engagement Increased susceptibility to attacks Insecure Retraining Pipelines Insufficient Logging Intellectual property theft Invalidated data Iteration and Improvement Lack of Compliance checks Lack of ethical guidelines Lack of performance monitoring Lack of robustness testing Legal fines Legal repercussions Limited Edge Case Handling Loan approval models were biased against minorities due to biased algorithm selection. Loss of proprietary information Misaligned Objectives Mitigation Mitigation Mitigations Mitigations Mitigations Mitigations Mitigations Mitigations Mitigations Model Deployment Model Drift Model drift in a predictive maintenance AI led to missed equipment failures Model Extraction Model Inversion Model Monitoring and Maintenance Model Optimization and Tuning Model Selection and Design Model Training and Evaluation Model Vulnerabilities Model watermarking Non-compliance Non-Compliance Operational failures Overfitting Performance degradation Poisoned data led to faulty medical diagnoses in a health AI system. Poor access controls Poor alignment with business needs Poor Governance Structures Poor logging of Anomalies Poor training data security Privacy breaches Privacy Breaches Problem Identification and Requirements Regular Audits Regular Security Audits Regulatory Evasion Techniques Regulatory fines Regulatory impact assessments Reliance on third-party data Reputation Damage Retraining schedules Risks Risks Risks Risks Risks Risks Risks Risks Risks Risks Robust version control Role-based access control Secure API management Secure Data Access Secure model environments Secure optimization practices Secure retraining environments, Security gaps Sensitive data was exposed during retraining of a customer service AI Stakeholder workshops Threats Threats Threats Threats Threats Threats Threats Threats Threats Threats Transparent AI usage policies Unaddressed Errors Unauthorized Access Unauthorized API Access Unauthorized use of the model Undefined Privacy Requirements Unexpected behaviors Unintentional Data Exposure Unprotected optimization processes Unreliable predictions Unsecured data handling Unsecured data sources Use of biased data Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Vulnerabilities Weak adversarial defenses Weak API security × delete