Discrepency on Usage Guidelines

I’ve recently read the updated Usage Guidelines and I’m confused between what I consider a discrepancy in technicality.

Under social media, under YouTube, it currently states:

“Disapproved use case example: writing a significant portion of a video’s script.”

However, just a small saunter above, under article writing, it remarks that:

“Writing assistance is one of the most powerful uses of GPT-3. We encourage use cases in this domain, for example copywriting and article writing.”

And goes on to remark:

“You’re welcome to create writing assistance applications, with the following restrictions:”

Given having the API assist in writing a video script is by definition writing assistance and basically article writing in all-but-name, that also includes user provided audio and video, effectively requiring a human-in-the-loop more so than article writing, why is in-the-loop article writing permitted, but all cases of video script writing not?

Could this rule be clarified/standardised? I feel like assistive script writing should be permissible so long as the scripts are manually reviewed and edited where appropriate using the same rules as article writing, assuming it’s in the same domain of overlap.

Could you reach out to support@openai.com - they’ll be able to clarify these guidelines for you.

It seems to me like a human-in-the-loop case should be allowed, but please don’t take my word for it.


Thanks a lot for bringing this to our attention, we really appreciate your feedback and encourage the discussion about these topics. With a human-in-the-loop, you can absolutely use the API for assistive script writing.

This is quite helpful as we work towards updating our guidelines to make them clearer & more consistent. Please don’t hesitate to reach out to us if you have any other questions.