Provides the reason for the action taken when harmful content is detected.
The input assessment.
The output from the model.
the output assessments.