1

Iā€™ve built an app that explains hard concepts in simple terms. ā€œExplain like Iā€™m 5ā€ type app.

Iā€™m using GPT4. And the app works like Google. You donā€™t need to sign up, simply ask a question and the app generates the answer.

Iā€™ve seen one or two queries for P*** (corn) and sex related questions. Just now somebody typed I want to kill myself. Not sure if it is real or he/she is just testing the app.

GPT4 handles this pretty nicely but since this is my API key OpenAI may see all these searches as me right? Iā€™m afraid I may get banned or smthā€¦ Iā€™d send userā€™s unique Ids but I donā€™t have users signed up.

Should I be worried?

2

You can use OpenAIā€™s free Moderation API to reduce the frequency of unsafe content sent to OpenAI.
Additionally, there are other helpful tips and advice provided in OpenAIā€™s safety best practices.

3 Likes
3

Hi @Bestbubbldev ,

unfortunately I think your gut feeling is correct and you should think about that.

What @whuwxl suggested is a good recommendation and you should absolutely implement that! To answer your question: You can visit the Usage policies (openai.com) and compare them to the information you submit.

An additional suggestion: You could additionally include a blacklist for potentially problematic terms and thus avoid sending the data over at all - this would be an additional step you can take. But I think moderation should work out fine :slight_smile:

What you told about your project sounds very helpful in the complex world we live in today and I wish you good luck with your endeavour!

1 Like