AAISM Practice Questions

A.  

Privilege escalation

B.  

Data poisoning

C.  

Model inversion

D.  

Evasion attack

A.  

Model stealing

B.  

Perfect knowledge

C.  

Data poisoning

D.  

Membership inference

A.  

Activate filtering logic to exclude intellectual property flags

B.  

Disclose service provider policies to declare compliance with regulations

C.  

Appoint a data steward specialized in AI to strengthen security governance

D.  

Review log information that records how data was collected

A.  

Risk tolerance

B.  

Risk threshold

C.  

Risk register

D.  

Risk appetite

A.  

Relying on human judgment for oversight

B.  

Limiting employee access to AI tools

C.  

Validating the provenance of the data source

D.  

Using a general-purpose large language model (LLM) to detect fraud

A.  

Establish an automated alert system for threshold breaches in risk metrics

B.  

Develop standardized risk reporting templates for different stakeholder groups

C.  

Implement real-time monitoring of key risk indicators (KRIs) for AI systems

D.  

Implement a risk dashboard for visualizing and tracking AI-related risk over time

A.  

Master services agreement

B.  

Shared responsibility model

C.  

Data minimization

D.  

Change management

A.  

The AI model prioritizes incidents based on business impact

B.  

AI is being used to monitor incident detection and alerts

C.  

The AI model’s outputs are validated by team members

D.  

Remediation of the AI system based on lessons learned

A.  

Collection and retention of only necessary data sets

B.  

Tracking and verification of data sets via cryptographic controls

C.  

Appropriate storage of data sets according to documented classification processes

D.  

Clear documentation of data sources, types used, and processing steps

A.  

Documented procedures for data sourcing, lineage tracking, and quality validation

B.  

Use of hashed identifiers to anonymize datasets used for model validation and internal analytics

C.  

Encrypted isolation and dynamic access controls on training data pipelines

D.  

Limitation of model training to structured data from vetted sources to minimize ingestion risk