We have covered sensitive data and content moderation. There is one other part of the platform that is not an endpoint protection, but instead the AI Firewall for Homegrown Applications.
Select the 'Homegrown Applications" from the menu on the left. Use the default policy and test in the playground. You will NOT be adjusting the policy here.
Exercise 4.1 - Jailbreak/Prompt Injection Protection
This is typically an API integration, but we can use the same playground for testing. Ignore the response from the internal mode, we are only testing the user input here.
Note that Prompt Injection/Jailbreak is a new protection not available in the Employee module.
Using public data sets or your own creativity, test this protection. Some samples are available under the ‘bug’ icon in the playground.
