I tried warning users here that it's not a good idea to use the outputs for commercial purposes like writing books.

Private company OpenAI, never know when or what they might do next with the generated text.

Well we must take into consideration how it may not be that simple to "watermark" generated text. Perhaps a encoding or hashing method will be needed...

We need something to do to GPT what Stable Diffusion did to DALL-E, so OpenAI doesn't have the opportunity to do stupid nanny state stuff like this anymore.

Nothing about this is "nanny state". It's a private corporation making a decision about their own product.

OpenAI should have done that before releasing it and they need to go further and actually create detectors for this. The same with Stable Diffusion and the rest of them.

They probably knew that they needed to urgently create detectors due to misuse, but still released it anyway.

Or we could all just stop worrying about who exactly wrote what.