If you’re constantly making something you’re afraid of and yet keep releasing it, you’re either a bullshitter (lying about it to hype it up), an imbecile (you could just stop doing the thing), deeply unethical (you won’t stop because personal gains matter more to you), or a mix of the above.

This is OpenAI's MO. Remember how they wouldn't release GPT-2 because of concerns of what people would do with it?

Though, I wonder how much of it is honest versus marketing. When I worked with the OpenAI team before the GPT-3 rollout, they had a lot of requirements in place about what we could and couldn't do. A lot of those requirements got in the way, so I imagine it wasn't all for show.

Am I the only one who thinks Sam is using these terms to gain some headlines attention? GPT-5’s experience has been horrible, and I’m not alone: https://news.ycombinator.com/item?id=44834777