This product provides an online testing feature. You can use this feature to test our efficient detection services for content compliance, sensitive content, and prompt attacks.
Prerequisites
Go to the AI Guardrails activation page to activate the AI Guardrails service.
Charges are incurred when you use the online testing feature. For more information, see Billing overview.
Procedure
Log on to the AI Guardrails console.
In the input box, enter the content that you want to test.
Below the input box, select a policy template for detection.
The following policy templates are available:
AI input content moderation (query_security_check_intl)
AI-generated content moderation (response_security_check_intl)
Click the Run test button.
View the Test results.
The following figure shows an example.
Alternatively, select a sample template. A set of detection templates is provided for content compliance, sensitive content, and prompt attacks. After you select a template, click the Run test button to view the detection results.
Enable features
After you run a test, if the status of Sensitive content detection or Prompt injection detection is Not enabled in the Test results, click Proceed to enable to activate the feature.
On the Check Item Configuration page, select the feature that you want to enable.
If you enable Sensitive content detection detection, take note of the message in the dialog box. This feature is billed separately. For more information, see Billing overview.
If you enable Prompt injection detection detection, take note of the message in the dialog box. This feature is in public preview and is available for a free trial. If you have any questions, contact your business manager.
