Using OpenAI to detect toxicity in comments?

I’m wondering if (and how) it would be possible to use OpenAI API to detect toxicity in a comment.

I have been eying perspectiveapi.com but it does not have swedish language support, but it seems that OpenAI understands swedish.

2 Likes

Welcome to the forum @arvidson

:wave: Hi!

A very cheap and easy solution could be to GOOGLE translate the comment and submit the result to the OpenAI toxicity engine.

Are you referring to OpenAI API är is there another toxicity engine? Or a way to create a prompt that finds toxicity in a comment?

My understanding is that the content-filter is a way to ensure that results coming from the API is not toxic.

Sorting out the false positives and investigating within the limits of the law and contractual agreements is a completely different story.

Yeah. I agree! :slight_smile: