SecAI+ Practice Test: Ready For The New AI Security Exam?

1. What does 'bias' in AI refer to?

Systematic prejudice in model predictions

High computational overhead

Slow data processing speed

Inadequate storage capacity

2. AI security governance includes oversight of:

Model development, deployment, and monitoring

Only hardware procurement

Only user interface design

Only marketing campaigns

3. Which practice helps prevent unauthorized access to AI models?

Implementing strong access controls and authentication

Publishing all code on public repositories

Removing all encryption protocols

Allowing unrestricted API access

4. What is the primary goal of AI threat modeling?

To identify and mitigate potential AI-specific risks

To maximize marketing reach

To reduce development time

To eliminate user support costs

5. The NIST framework emphasizes which core function for AI security?

Map, Measure, Manage, and Govern

Delete, Encrypt, Backup, Archive

Install, Update, Patch, Reboot

Scan, Block, Report, Isolate

6. Which is a critical control for AI model monitoring?

Detecting performance degradation and anomalies

Using only open-source frameworks

Avoiding all logging and auditing

Disabling user authentication

7. What is a backdoor attack in AI systems?

Hidden malicious behavior triggered by specific inputs

Physical access to server rooms

Weak password policies only

Outdated software libraries

8. AI model transparency is important for:

Accountability and regulatory compliance

Increasing processing speed only

Reducing software licensing costs

Eliminating the need for testing

9. Which attack targets the output of an AI model to cause harm?

Evasion attack

Privilege escalation

SQL injection

Man-in-the-middle attack

10. Federated learning enhances security by:

Training models on decentralized data without centralizing it

Eliminating the need for encryption

Removing all user privacy controls

Increasing data collection on users

11. What is prompt injection in AI security?

Injecting malicious code into system memory

Manipulating AI model inputs to produce unintended outputs

Inserting SQL commands into a database query

Physically tampering with server hardware

12. Which of the following is a key component of AI supply chain security?

Vetting third-party models and dependencies

Installing antivirus software only

Using only proprietary hardware

Disabling all network connections

13. What is differential privacy in AI systems?

A method to encrypt network traffic

A technique to protect individual privacy in datasets

A firewall configuration standard

A backup and recovery strategy

14. Adversarial examples in AI security are inputs designed to:

Improve model training efficiency

Fool or mislead AI models into wrong predictions

Reduce memory consumption

Enhance user interface responsiveness

15. Which threat involves stealing trained AI models for unauthorized use?

Model extraction attacks

Cross-site scripting

Buffer overflow exploits

Denial-of-service flooding

16. What is the primary purpose of model explainability (XAI)?

To increase computational speed

To understand how models make decisions

To reduce data storage costs

To bypass authentication mechanisms

17. Data poisoning attacks target which component of an AI system?

The training dataset

The user's browser cache

Network firewall rules

Operating system kernel

18. Which framework is commonly used for AI governance and risk management?

NIST AI Risk Management Framework

Windows Defender Policy

Apache HTTP Server Config

Docker Container Orchestration

19. What does 'hallucination' mean in AI context?

Visual distortions in rendered images

AI generating false or fabricated information confidently

Errors caused by overheating GPUs

User interface glitches during operation

20. Which of the following is a primary risk of model poisoning?

Models generate toxic or biased outputs

Network bandwidth is consumed excessively

User passwords are compromised

Server CPU temperature increases

SecAI+ Practice Test: Ready for the New AI Security Exam?