according to open AI TOS you can send them an email to opt out of them using your prompts for model training because the right to the "content" you send to their API is perpetually yours. they also license the outputs to you to use as see fit (since you're paying them to use the service). so, on paper, you can use it for "serious" purposes without an issue. that said, the legal aspects of this whole thing is still work in progress and I strongly suggest you discuss the implications with your internal legal team before sending any sensitive company info