---
sourceDocument: Yokohama Enable AI
sourceDocumentLink: https://servicenow-prod.fluidtopics.net/r/yokohama/intelligent-experiences

 Release :

    - yokohama

ft:locale :

    - en-US

ft:publication_title :

    - Yokohama Enable AI

ft:clusterId :

    - platai

bundleId :

    - platai

workflow :

    - Platform


---

# Configure sensitive topic filters

# Configure sensitive topic filters {#ariaid-title1}

Release version: Yokohama  
Updated July 31, 2025  
![](https://www.servicenow.com/docs/portal-asset/ico-clock) 2 minutes to read  
Set up filters in AI Guardian to redirect Virtual Agent users to a live agent or HR case when a sensitive subject is detected in a conversation.

## Before you begin

Role required: sn_nowassist_admin.nsa_admin

## About this task

AI Guardian detects sensitive subjects in Virtual Agent conversations and redirect users to the Sensitivity Detection: Fallback Virtual Agent topic instead of continuing the conversation with generative AI. You can activate and configure the subjects that trigger this redirection, and add sample phrases to improve detection accuracy.  
Note:  
The sensitive topic filter guardrail is available only with HR Service Delivery and applies only to Virtual Agent conversational skills.

See [Now Assist Guardian](https://servicenow-prod.fluidtopics.net/H0bVbM_NdghPjzLAF~OnJg#now-assist-guardian "AI Guardian is built on the ServiceNow Small Language Model (SLM) and monitors generative AI interactions to detect offensive content, prompt injection attacks, and sensitive topics.") for more information.

## Procedure

1. Navigate to AllNow Assist AdminSettings.
2. Navigate to Now Assist GuardianFilters.
3. Go to the Available for you tab to see which filters are available.
4. Select Activate on the filter you want to enable.
5. Review and generate sample phrases.  
   Sample phrases tell the LLM what kind of text the filter should detect. The more sample phrases you provide, the more accurately the LLM detects these topics.

   | Option | Description |
   | Add a phrase | 1. Select New phrase to add a phrase manually. 2. Enter text in the Add new phrase text box. 3. Select Save. {#configure-sensitive-topic-filters__ol_q5k_pjt_q3c} |
   | Generate a phrase using generative AI | 1. In the Generative samples panel, select the maximum number of phrases to generate from the list. 2. Select Generate. 3. Select the check box next to the phrase and selectAdd to add individual sample phrase. 4. Select the check box next to each phrase and then selectAdd all to add multiple phrases together. {#configure-sensitive-topic-filters__ol_r5k_pjt_q3c} |
   | Edit a phrase | 1. Select the pencil icon and start editing the phrase. 2. select Save. {#configure-sensitive-topic-filters__ol_s5k_pjt_q3c} |
   | Delete a phrase | Select the delete icon to remove the phrase. |
   |-|-|

   {#configure-sensitive-topic-filters__choicetable_p5k_pjt_q3c}
6. Select Save and Continue.
7. Choose which Virtual Agents you want to enable the sensitive topic filters for by selecting the checkbox next to the Virtual Agent name.
8. Select Save and Continue.
9. Review the fallback topic for when a filter is detected.  
   By default, when a sensitive topic is detected, the user is redirected to the Sensitivity Detection: Fallback topic, which prompts the user to create an HR case or speak to a live agent. You can preview the topic from the guided
   setup.
10. Select Save and continue.
11. Review your filter configuration in the final step.  
    If you want to make any changes, use the step navigation panel to return to a previous step.
12. Select Activate to activate the filter.

## Result

The sensitive topic filter is activated and appears in the Active tab of the Filters section of AI Guardian settings. When the filter detects a sensitive topic in a Virtual Agent conversation, the user is redirected to the Sensitivity Detection: Fallback Virtual Agent topic.

## What to do next

To modify the filter configuration after activation, go to the Active tab of the AI Guardian settings options and select the more actions icon (![More options icon.]()) in the Actions column.

*[\>]: and then