Question

A retail company uses a foundation model on Amazon Bedrock to generate product descriptions. Marketing wants every generated description to avoid mentioning competitor brand names, specific discount percentages, or medical claims. The team wants to control this behavior directly through how they write the prompt, without retraining or fine-tuning the model. Which prompt engineering approach best achieves this requirement?

Accepted Answer

Add explicit instructions in the prompt telling the model what topics and terms to exclude from its output (negative prompting).

Answer

Increase the temperature parameter so the model produces more varied text that naturally avoids restricted terms.

Answer

Reduce the maximum token count so the response is too short to include restricted content.

Answer

Fine-tune the foundation model on a dataset that removes all restricted terms.