Viewing AI Guardrails Incidents

AI guardrails incidents is a consolidated view of Skope IT alerts. It pulls alerts containing Alert Type = AI Security. Also, if a single prompt or response hits multiple AI Guardrails profiles, it generates multiple alerts. For example, if you have a profile that detects the Weapons predefined category and another profile that detects a key word, such as “gun”, then a prompt like “How to buy a gun.” will generate two alerts. However, all alerts for single prompt are aggregated into a single instance, so the prompt or response will only have one logged incident.

On the AI Guardrails Incidents page (Incidents > AI Guardrails Incidents), you can:

1. Filter the AI guardrails incident list using the following options:

   • Time: Filter incidents by a specific time frame. You can use a predefined time frame or choose Date Range to use the calendar and time menus to customize your own.

   • Filters: Click + Add Filter to filter the incidents and narrow your search results. You also can click  to search and filter the incidents by a query.

2. View a list of AI guardrails incidents. For each incident, you can see the following information:

   • Incident ID: The ID of the incident.

   • Activity: The type of activity performed with the Gen AI app. To learn more:

   • Category: The matched content category of the prompt or response. To learn more: AI Guardrails Profile.

   • User: The user that triggered the incident.

   • Action: The action taken based on your policy.

   • Application: The Gen AI application used.

   • Conversation ID: The ID of the conversation with the Gen AI app.

   • Detection Time: The time Netskope detected the incident.

3. Click to view the incident details where you can see the matched content that triggered the incident. You can also click Report False Positive if you believe this incident is a false positive.

   

4. View multiple pages of the table.

5. View up to 100 sites per page.

You can also leverage the AI Guardrails Dashboard within Netskope Advanced Analytics, which gives security teams visibility into prompts and responses within Gen AI apps across the enterprise and how well AI-specific security policies are performing.