IBM Cloud Docs
Hate, abuse, and profanity filter

Hate, abuse, and profanity filter

You can use the hate, abuse, and profanity (HAP) filter to maintain an inclusive environment. This feature filters content to provide a positive online atmosphere and a safe community for your customers.

By enabling HAP detection, you create a more consistent and reliable user experience, safeguarding against the rare instances where HAP content might be generated.

HAP filter
HAP filter

Enabling the HAP filter

To enable the HAP filter:

  1. Go to Global settings > Generative AI > HAP filtering
  2. Click the toggle to turn it on.

The HAP filter applies to the following areas:

  • Conversational search
  • Content-grounded answering
  • General-purpose answering
  • AI-guided actions

Consider the following key points when you use the HAP filter:

HAP detection: If HAP is detected, the assistant provides a generic fallback response. For more information, see Conversational search.

Recommendation: When HAP is on, it minimizes the risk of unintended or inappropriate responses.