AI System Security Lifecycle Quiz

1. What does model drift refer to?

Gradual degradation of model performance over time

Physical movement of servers

Unintended bias in predictions

Increase in model size

2. Bias in AI systems can introduce security vulnerabilities and discriminatory outcomes. True or False?

True

False

3. Which of the following is essential for maintaining AI system security post-deployment?

Regular security audits and vulnerability assessments

Disabling all monitoring

Freezing the model indefinitely

Removing access logs

Submit

5. What is the primary benefit of using federated learning in AI security?

Faster model training

Training models without centralizing sensitive data

Reducing computational costs

Improving model accuracy

6. Security patches for AI systems should be deployed immediately regardless of testing impact. True or False?

True

False

7. Which security control prevents unauthorized access to model artifacts and training data?

Logging and monitoring

Access control and encryption

Performance tuning

Model compression

Submit

9. In secure AI development, what is the purpose of code review?

To increase development speed

Detect vulnerabilities and ensure security best practices

To reduce model accuracy

To eliminate testing requirements

10. Differential privacy protects individual records by adding controlled noise to data. True or False?

True

False

11. What is the primary goal of threat modeling in AI systems?

Identify potential vulnerabilities and attack vectors before deployment

Document all software dependencies

Establish employee access controls

Monitor real-time system performance metrics

12. Continuous monitoring of AI systems in production is optional if the model was thoroughly tested. True or False?

True

False

13. Which attack targets the confidentiality of training data by extracting it from a trained model?

Evasion attack

Membership inference attack

Backdoor attack

Byzantine attack

Submit

15. API security in AI systems should include rate limiting and input validation. True or False?

True

False

16. Which of the following is a key component of secure model deployment?

Using public cloud infrastructure without encryption

Containerization with runtime security monitoring

Storing credentials in source code

Disabling audit logging

Submit

18. What does GDPR require regarding personal data used in AI training?

Encryption of all training data

Consent, transparency, and right to deletion

Weekly security audits

Use of public datasets only

19. Adversarial examples are inputs designed to cause a model to make incorrect predictions. True or False?

True

False

20. Which phase of the AI lifecycle focuses on identifying and mitigating data poisoning risks?

Deployment

Training and validation

Monitoring and maintenance

Model retirement

SecAI+ AI System Security Lifecycle Overview Quiz

1. What does model drift refer to?

2.

What first name or nickname would you like us to use?

2. Bias in AI systems can introduce security vulnerabilities and discriminatory outcomes. True or False?

3. Which of the following is essential for maintaining AI system security post-deployment?

4. Model ______ involves using one trained model to improve another model's performance.

5. What is the primary benefit of using federated learning in AI security?

6. Security patches for AI systems should be deployed immediately regardless of testing impact. True or False?

7. Which security control prevents unauthorized access to model artifacts and training data?

8. A ______ attack injects malicious data during training to compromise model behavior.

9. In secure AI development, what is the purpose of code review?

10. Differential privacy protects individual records by adding controlled noise to data. True or False?

11. What is the primary goal of threat modeling in AI systems?

12. Continuous monitoring of AI systems in production is optional if the model was thoroughly tested. True or False?

13. Which attack targets the confidentiality of training data by extracting it from a trained model?

14. What is model ______ and why is it important in AI security?

15. API security in AI systems should include rate limiting and input validation. True or False?

16. Which of the following is a key component of secure model deployment?

17. Model ______ refers to the process of verifying a model's performance on unseen data.

18. What does GDPR require regarding personal data used in AI training?

19. Adversarial examples are inputs designed to cause a model to make incorrect predictions. True or False?

20. Which phase of the AI lifecycle focuses on identifying and mitigating data poisoning risks?