MuleSoft inference Connector - [Toxicity] Detection by Text

About this lesson

Customer-facing channels (chat widget, social posts, support comments) sometimes contain abusive, harassing, hateful, or threatening language. [Toxicity] Detection by Text operation to interact with OpenAI and This operation is used to detect harmful content in a text. You need an automated, reliable filter that: Detects toxicity (and subtypes: insult, hate, sexual content, threats, self-harm, etc.) in real time, Scores severity/confidence, Applies business rules (auto-moderate, flag for human review, throttle user, escalate to safety team), Preserves auditability and privacy.

Original Description