AAISM Practice Test 14 — 15 Questions

Question 1

An organization is implementing 'Red Teaming' for AI. What is the main governance objective?

Accepted Answer

To proactively identify vulnerabilities and failure modes.. Red teaming for AI is a structured adversarial exercise where internal or external experts deliberately attempt to make the model fail — via prompt injection, jailbreaks, data poisoning, bias elicitation, or tool misuse — to surface vulnerabilities before real attackers do. The governance objective is proactive risk identification: producing a documented inventory of failure modes, their severity, and recommended mitigations that feed into the organization's risk register and deployment decisions.

Answer

To reduce operational costs.

Answer

To speed up deployment.

Answer

To increase the model size.

Question 2

Which TWO of the following are recommended when auditing an 'AI Model' for bias and fairness?

Accepted Answer

Measure performance metrics across subgroups. Option A is correct because auditing an AI model for bias and fairness requires disaggregating performance metrics (e.g., accuracy, precision, recall, false positive/negative rates) across protected or relevant subgroups to detect disparate impact that aggregate metrics would hide. Option E is correct because bias often originates in the training data, so analyzing it for representational imbalances (under/over-representation of groups, label bias, sampling skew) is a foundational fairness audit step. Option B is wrong because removing all logging destroys the audit trail and traceability needed for accountability, and anonymity is not achieved by eliminating logs. Option C is wrong because adding layers changes model capacity, not fairness, and can worsen overfitting or bias. Option D is wrong because deploying a model for all possible use cases ignores context-specific validity and increases the risk of harmful, out-of-scope bias.

Answer

Remove all logging to maintain anonymity

Answer

Increase the number of layers in the model

Answer

Use the model for all possible use cases

Question 3

A company is adopting the ISO/IEC 42001 AI management system standard. What is the primary purpose of the 'Context of the Organization' clause?

Accepted Answer

To define the boundaries and strategic intent of the AI management system.. In ISO/IEC 42001, Clause 4 'Context of the Organization' establishes why the AI management system exists and what it must cover. It requires the organization to determine external and internal issues, identify interested parties and their requirements, and define the scope and boundaries of the AIMS aligned with strategic direction. Option A captures this purpose: defining boundaries and strategic intent so all subsequent clauses (leadership, planning, support, operation, evaluation, improvement) operate within a coherent frame.

Answer

To specify hardware requirements for training.

Answer

To mandate the use of cloud-based AI services.

Answer

To identify all bugs in the production code.

Question 4

When assessing AI third-party vendor risk, which document is most useful for understanding the vendor's data handling practices?

Accepted Answer

The Data Processing Agreement (DPA).. A Data Processing Agreement (DPA) is the contractual document that specifies how a vendor processes, stores, transfers, and protects personal data, including subprocessors, retention, breach notification, and cross-border transfer mechanisms. It is the primary artifact for assessing third-party data handling risk under regulations like GDPR and CCPA.

Answer

The user interface mockups.

Answer

The company's marketing brochure.

Answer

The vendor's quarterly earnings statement.

Question 5

Which THREE security controls are effective against 'Data Leakage' in AI?

Accepted Answer

Data Loss Prevention (DLP) tools.. Data Loss Prevention (DLP) tools (A) are effective because they inspect and block sensitive data such as PII, PHI, or credentials from leaving approved boundaries via email, endpoints, or cloud uploads, directly mitigating exfiltration. Granular Role-Based Access Control (RBAC) (C) limits which users, service accounts, or pipelines can query or retrieve sensitive datasets and model outputs, enforcing least privilege so unauthorized parties cannot access data that could leak. AI output filtering for PII (D) scans model responses before delivery and redacts or blocks regulated identifiers, preventing leakage through generated text that may reproduce memorized or inferred sensitive data. A louder cooling system (B) is purely a physical/thermal concern and has no bearing on confidentiality or data exfiltration. Increasing the model's training duration (E) affects model quality or convergence, not leakage prevention, and could even increase memorization risk.

Answer

Using a louder cooling system.

Answer

Increasing the model's training duration.

Question 6

Which TWO measures are most effective in reducing the risk of 'Hallucinations' in LLM systems?

Accepted Answer

Using Retrieval-Augmented Generation (RAG).. Option B, Retrieval-Augmented Generation (RAG), is correct because it grounds the model's responses in retrieved, verifiable external documents at inference time, so the LLM conditions its output on factual context rather than relying solely on parametric memory, which substantially reduces fabricated content. Option C, providing clear system prompts defining constraints, is correct because explicit instructions (e.g., 'answer only from the provided context,' 'say I don't know if unsure') steer the model away from speculative generation and reduce unsupported claims. Option A is not effective because simply training on larger public internet datasets does not eliminate hallucination and can even amplify it, since web data contains misinformation and the model still lacks a grounding mechanism at inference. Option D is irrelevant because removing punctuation does not affect factual grounding and may degrade input comprehension. Option E is counterproductive because raising temperature to 2.0 increases sampling randomness and diversity, which typically increases hallucination rather than reducing it.

Answer

Training the model on larger public internet datasets.

Answer

Removing all punctuation from the input.

Answer

Increasing the model's temperature parameter to 2.0.

Question 7

In the context of poisoning attacks, which technique is most effective for detecting data contamination in an automated ML training pipeline?

Accepted Answer

Statistical outlier detection on training feature sets. Statistical outlier detection on training feature sets is the most effective technique because poisoning attacks introduce anomalous data points that deviate from the expected distribution of legitimate training data. By applying statistical methods (e.g., z-score, IQR, isolation forests) to feature distributions, contaminated samples can be flagged and removed before they influence model weights. This directly targets the integrity of the training pipeline at the data ingestion stage.

Answer

Increasing training epochs

Answer

Differential Privacy

Answer

Model quantization

Question 8

What is 'AI Red Teaming'?

Accepted Answer

Proactive adversarial testing to identify model weaknesses.. AI Red Teaming is the practice of proactively and adversarially probing an AI system to uncover weaknesses such as prompt injection, jailbreaks, bias, data leakage, and unsafe outputs before malicious actors exploit them. It mirrors traditional security red teaming but targets model behavior, guardrails, and the surrounding application stack. The goal is to find and remediate vulnerabilities through structured adversarial testing.

Answer

The marketing department's review of AI ads.

Answer

An automated script that changes the model color.

Answer

The process of painting red marks on the server racks.

Question 9

What is 'AI Risk Appetite'?

Accepted Answer

The level of risk an organization is willing to accept.. AI Risk Appetite is the amount and type of risk an organization is willing to accept in pursuit of its AI objectives — it's a governance statement that bounds decision-making. It translates board-level tolerance into operational thresholds that guide model deployment, monitoring, and escalation. Without a defined appetite, risk decisions become ad hoc and inconsistent.

Answer

The amount of money spent on GPUs.

Answer

The number of employees using AI.

Answer

The speed of the AI model.

Question 10

Which of the following describes the purpose of 'Differential Privacy' in a machine learning system?

Accepted Answer

To prevent the identification of individuals in the training data. Differential Privacy (DP) is a mathematical framework that adds calibrated noise to data or computations so that the inclusion or exclusion of any single individual's data has a negligible effect on the output. In ML, it prevents attackers from determining whether a specific person was in the training set or inferring their attributes, thereby protecting individual privacy while still allowing aggregate statistical learning.

Answer

To enable cross-cloud model training

Answer

To allow for better interpretation of model layers

Answer

To ensure the model is faster at inference time

Question 11

A company is performing an 'AI Impact Assessment' (AIA). Which domain is the most critical for identifying 'Privacy' risks?

Accepted Answer

Data lineage, sensitivity, and processing purpose.. Privacy risk in an AI Impact Assessment is fundamentally about data: where it came from (lineage), how sensitive it is (PII, PHI, financial), and why it is being processed (purpose limitation). These three elements determine whether the processing complies with GDPR, CCPA, and similar regulations, and whether privacy harms (re-identification, inference, secondary use) are possible. Without this understanding, no meaningful privacy risk assessment can occur.

Answer

Model training costs.

Answer

UI design features.

Answer

Server CPU utilization.

Question 12

You are configuring Microsoft Defender for Cloud to monitor your Azure OpenAI service. Which specific resource diagnostic setting must be enabled to ensure that prompt injection attempts are logged in the Log Analytics workspace?

Accepted Answer

Enable 'RequestResponse' diagnostic logs for the AI resource. To log prompt injection attempts against Azure OpenAI in Microsoft Defender for Cloud, the 'RequestResponse' diagnostic logs must be enabled for the AI resource, as these capture the actual prompts and completions where injection patterns appear. This setting routes request and response data to the Log Analytics workspace, enabling Defender for Cloud's AI threat protection to detect and alert on prompt injection. Other logging categories do not capture the prompt-level content needed for this detection.

Answer

Enable 'AccessControl' logging in Entra ID

Answer

Enable 'MetricAlerts' at the subscription level

Answer

Enable 'AuditLogs' in the Storage Account settings

Question 13

What is the primary role of an 'AI Ethics Board' in an organization's governance structure?

Accepted Answer

Reviewing AI use cases against organizational values and principles.. An AI Ethics Board is a governance body whose primary function is to review AI use cases, projects, and deployments against the organization's stated ethical values, principles, and policies. It provides oversight and guidance on fairness, transparency, accountability, and societal impact — not operational execution. This aligns with governance frameworks like the NIST AI RMF 'Govern' function and ISO/IEC 42001.

Answer

Directing daily stock market trades.

Answer

Conducting server hardware maintenance.

Answer

Writing all the Python code for the AI.

Question 14

Which department should own the 'AI Ethics Policy'?

Accepted Answer

Compliance/Legal or Ethics Office.. The AI Ethics Policy sets normative expectations for how AI is developed and used, covering fairness, transparency, privacy, and accountability. Ownership belongs with Compliance/Legal or the Ethics Office (D) because these functions interpret regulatory obligations, manage ethical risk, and have the independence and authority to enforce policy across business units. Placing ownership there ensures the policy is defensible, aligned with law and regulation, and not subordinated to IT operational priorities.

Answer

The IT helpdesk.

Answer

The cafeteria management.

Answer

The building security team.

Question 15

What is the goal of 'Data Sanitization' before training?

Accepted Answer

Remove PII and malicious data from the training set. Data sanitization is the process of cleaning a training dataset by removing personally identifiable information (PII), toxic content, and malicious or poisoned samples before the data is used to train a model. This reduces privacy risk, regulatory exposure, and the chance that an attacker can manipulate model behavior through data poisoning. The goal is not to change formats or reduce costs but to ensure the training data is safe and compliant.

Answer

Change the file format of the dataset

Answer

Reduce the cloud storage costs

Answer

Increase the number of images in the dataset