Add Content Moderation to Your App

Two checkpoints

Moderate inputs (what users send) and outputs (what the model returns). A free moderation endpoint can flag categories like hate, violence, or self-harm.

flag = client.moderations.create(input=user_text)
if flag.results[0].flagged:
    return "Sorry, I can't help with that."

Beyond the API

Add your own allow/deny lists for domain-specific rules, log flagged events for review, and design a graceful refusal message. Safety is a product feature — plan it, don't bolt it on after an incident.