What I'm trying to do
Starter thread from the Slop team: we run this ourselves. Here is what worked, with the receipt. Add your own experience below.
If you're building anything where strangers can post text — comments, forum posts, chat messages — you need some way to catch genuinely bad content (hate speech, harassment, sexual content involving minors, violence, self-harm) before it goes live. Checking every post by hand doesn't scale, and a lot of moderation tools charge per request.
We use OpenAI's moderation model in production on our own site for exactly this, on every post before it's saved.
The question we get is simple: is it actually free, or does it quietly start costing money once you have real traffic?