AWS Certified AI Practitioner Practice Exam

What is the recommended approach to prevent a disease detection model from revealing personal patient information?

1. Use Amazon Macie to scan the model's output for sensitive data and set up alerts

2. Configure AWS CloudTrail to monitor the model's responses and create alerts

3. Use Guardrails for Amazon Bedrock to filter content and set up alarms for violations

4. Implement Amazon SageMaker Model Monitor to detect data drift

The correct answer is: Use Guardrails for Amazon Bedrock to filter content and set up alarms for violations

The recommended approach to prevent a disease detection model from revealing personal patient information is to use Guardrails for Amazon Bedrock to filter content and set up alarms for violations. This option focuses on creating a safety net that ensures the information generated or inferred by the model adheres to compliance standards and privacy regulations. Guardrails help in establishing content moderation layers that can prevent sensitive information from being disclosed inadvertently. By implementing such guardrails, the system can actively filter out any content that may include personal patient data, minimizing the risk of privacy violations. Additionally, the ability to set up alarms for violations ensures that any breaches or potential leaks are promptly identified and addressed, enhancing overall data governance. While scanning model outputs, monitoring responses, and detecting data drift are important practices, they may not be tailored specifically to prevent the exposure of personal information in the same direct, preventive manner as guardrails do. Guardrails provide proactive measures specifically designed to address data privacy concerns, making them the most effective choice in this scenario.