A generative AI system produces different outputs for the same input on consecutive calls. This property is called:
#61
Answer: D✅ Correct❌ Incorrect
Model drift occurs when a deployed model's performance degrades over time. Which type of drift is caused by changes in the statistical properties of input features?
#62
Answer: C✅ Correct❌ Incorrect
ISO/IEC 42001 Annex A controls specifically address which AI-related concern?
#63
Answer: D✅ Correct❌ Incorrect
Which emerging AI threat involves attackers injecting hidden instructions into content that an AI agent will process (e.g., a web page or email), causing the agent to perform unintended actions?
#64
Answer: A✅ Correct❌ Incorrect
Which technique can help prevent sensitive training data from being reconstructed from a deployed model's outputs?
#65
Answer: C✅ Correct❌ Incorrect
A 'Trojan' or backdoor attack on a machine learning model is characterized by:
#66
Answer: C✅ Correct❌ Incorrect
Which practice is part of responsible AI lifecycle management at the model retirement stage?
#67
Answer: D✅ Correct❌ Incorrect
The NIST AI Risk Management Framework (AI RMF) organizes its guidance into four core functions. Which of the following is NOT one of them?
#68
Answer: B✅ Correct❌ Incorrect
What is 'model versioning' and why is it important in the AI lifecycle?
#69
Answer: A✅ Correct❌ Incorrect
Which control reduces the risk that an AI agent with tool-use capabilities (e.g., code execution, web browsing) will perform irreversible harmful actions?
#70
Answer: B✅ Correct❌ Incorrect
Under GDPR, using personal data to train an AI model without a valid legal basis constitutes:
#71
Answer: B✅ Correct❌ Incorrect
Which of the following is a key risk associated with using public generative AI tools (e.g., consumer ChatGPT) for processing confidential business documents?
#72
Answer: D✅ Correct❌ Incorrect
Which property of AI systems ensures that accountability can be traced when an AI system causes harm?
#73
Answer: C✅ Correct❌ Incorrect
Which of the following frameworks explicitly addresses the trustworthiness dimensions of AI systems including reliability, safety, security, privacy, fairness, and accountability?
#74
Answer: C✅ Correct❌ Incorrect
An adversarial example is best defined as:
#75
Answer: C✅ Correct❌ Incorrect
Which of the following cloud security controls is MOST important for protecting LLM API keys used by production applications?
#76
Answer: B✅ Correct❌ Incorrect
Which of the following access control principles is MOST important when granting AI systems access to enterprise APIs and databases?
#77
Answer: A✅ Correct❌ Incorrect
Which of the following best describes a Large Language Model (LLM)?
#78
Answer: B✅ Correct❌ Incorrect
An organization wants to demonstrate that its AI system complies with ISO 42001. What is the most appropriate approach?
#79
Answer: C✅ Correct❌ Incorrect
Which MITRE ATLAS technique describes an attacker using the target model's own predictions to infer properties of its training data?
#80
Answer: B✅ Correct❌ Incorrect
Which of the following best characterizes the relationship between AI safety and AI security?
#81
Answer: D✅ Correct❌ Incorrect
Which of the following describes a 'data sheet for datasets' (Datasheets for Datasets) in the context of AI transparency?
#82
Answer: A✅ Correct❌ Incorrect
The AI supply chain risk of using a maliciously modified open-source ML library (e.g., a compromised version of NumPy or PyTorch) is best mitigated by:
#83
Answer: C✅ Correct❌ Incorrect
What does 'grounding' an LLM mean in the context of enterprise AI deployments?
#84
Answer: B✅ Correct❌ Incorrect
Which of the following best describes 'data lineage' in the context of AI data governance?
#85
Answer: C✅ Correct❌ Incorrect
Which document published by CSA specifically maps AI control objectives to ISO 42001, ISO 27001, NIST AI RMF, and BSI AIC4?
#86
Answer: C✅ Correct❌ Incorrect
Which of the following is the most direct consequence of inadequate access controls on an AI model's training data pipeline?
#87
Answer: B✅ Correct❌ Incorrect
Which security control is most effective at protecting sensitive data used to train AI models stored in a cloud object store?
#88
Answer: D✅ Correct❌ Incorrect
Which of the following AI risk assessment activities aligns with the NIST AI RMF 'MAP' function?
#89
Answer: D✅ Correct❌ Incorrect
Which AI safety concept describes the property that an AI system should behave as intended across diverse, edge-case, and adversarial conditions?