AWS Certified AI Practitioner (AIF-C01) — Question 324

A company is building a new generative AI chatbot. The chatbot uses an Amazon Bedrock foundation model (FM) to generate responses. During testing, the company notices that the chatbot is prone to prompt injection attacks.

What can the company do to secure the chatbot with the LEAST implementation effort?

Answer options

• A. Fine-tune the FM to avoid harmful responses.
• B. Use Amazon Bedrock Guardrails content filters and denied topics.
• C. Change the FM to a more secure FM.
• D. Use chain-of-thought prompting to produce secure responses.

Correct answer: B

Explanation

Amazon Bedrock Guardrails provides built-in content filtering and denied topics capabilities specifically designed to mitigate prompt injection and jailbreak attempts with minimal configuration effort. Fine-tuning an FM or switching models requires significant time, data, and computational resources, making them high-effort solutions. Chain-of-thought prompting is a prompt engineering technique designed to improve reasoning, not a reliable security mechanism to prevent malicious injections.